Research ArticleAn Improved Route-Finding Algorithm Using UbiquitousOntology-Based Experiences Modeling

Maryam Barzegar1 Abolghasem Sadeghi-Niaraki 23 Maryam Shakeri2

and Soo-Mi Choi 3

1Centre for Spatial Data Infrastructures and Land Administration Department of Infrastructure Engineeringe University of Melbourne Melbourne Australia2Geoinformation Technology Center of Excellence Faculty of Geodesy amp Geomatics EngineeringK N Toosi University of Technology Tehran Iran3Department of Computer Science and Engineering Sejong University Seoul Republic of Korea

Correspondence should be addressed to Abolghasem Sadeghi-Niaraki asadeqi313gmailcom

Maryam Barzegar and Abolghasem Sadeghi-Niaraki contributed equally to this work

Received 9 April 2019 Revised 9 August 2019 Accepted 16 September 2019 Published 11 November 2019

Academic Editor Mahdi Jalili

Copyright copy 2019 Maryam Barzegar et al is is an open access article distributed under the Creative Commons AttributionLicense which permits unrestricted use distribution and reproduction in any medium provided the original work isproperly cited

Every day people are hired in dierent organizations and old and retiring employees are eliminated from enterprise systems Eliminatingthese individuals from organizations leads to the loss of their spatial experiences In addition since new employees lack relevant ex-perience they need a long time to develop the correct skills for the company and may even cause damage to the organization during thislearning processerefore storing the spatial experience of individuals is a critical issue Due to the intelligence of ubiquitous GeospatialInformation System (GIS) any experience from any user can be received and stored In the future based on these experiences anappropriate service to each user may be provided as needed is paper aims to propose an ontology-based model to store spatialexperiences in the eld of ubiquitousGIS route nding For this purpose rst ontology is designed for route nding and then according tothis ontology an ontology-based route-nding algorithm is developed for ubiquitous GIS Finally this algorithm is implemented forTehran Iran and its results are compared with the shortest path algorithm (Dijkstrarsquos algorithm) in terms of the route length and traveltime for peak trac timee results show that while the route length obtained from the ontology-based algorithm ismore thanDijkstrarsquosalgorithm the travel time is lower and on some routes the dierence in travel time saved reaches 35 minutes

1 Introduction

Knowledge management is a process that helps organiza-tions to detect select organize and publish important in-formation and skills that are considered as organizationalmemory and which are not typically organized is enablesorganizations to eectively manage learning issues strategicplanning and dynamic decision making e causes of theadvent of knowledge management can be considered asfollows (1) transformation of the industrial business modelin the past the assets of an organization were fundamentallytangible and nancial assets (production facilities cars landand so on) (2) an extraordinary increase in the amount of

information and its electronic storage and increased accessto information have generally added value to knowledge (3)the changes in the age pyramid of populations and thedemographic properties that are mentioned in only a fewsources and (4) specializing in activities may also hold therisk of losing organizational and expertise knowledgethrough the transfer or dismissal of employees Taxi asso-ciations are one of the organizations that need to organizeorganizational memory In this organization taxi driversacquire skills and knowledge by repeating their daily routemany times so that after some years they become expertsand experienced person in their work Actually this expe-rience is a kind of knowledge which is gained after many

HindawiComplexityVolume 2019 Article ID 9584397 15 pageshttpsdoiorg10115520199584397

years of effort and it is called spatial experience because it isbased on location A spatial experience is an everyday expe-rience of the population that is related to locations or peoplersquosactivities that are conducted in a location [1] In other wordsspatial experience can be called location-based experience Asfinding a route between locations and destinations by taxidrivers is a location-based activity that is conducted by taxidrivers in a city the experiences of taxi drivers in finding goodroutes are spatial experiences On a daily basis there are manyexperienced taxi drivers who are removed from taxi associ-ations after they retire -is causes a loss of the spatial ex-periences of expert drivers without transferring thoseexperiences to the next driver On the other hand new taxidrivers need high-cost training and considerable time to gainexperience comparable to that of the previous drivers Due tothe aforementioned reasons it is necessary to usemethods thatcan save this knowledge and reuse it to solve various problemsSemantics plays a significant role in knowledge organizationand it can support the enrichment of measurements andgaining knowledge [2] Ontology is one of themethods that usesemantics Ontology is a collection of terms in a domain andthese terms are linked to visual properties relationships andassociations Ontology structures domain knowledge providesthe opportunity of reusing domain knowledge and makesdomain assumptions explicit [3]

Several researchers have focused on the storage of expe-riences and formulation of knowledge Foguem et al [4]formulated knowledge in the feedback processes of experience-eir purpose is to transform the information gained from theexperience into explicit knowledge -e feedback frame ofexperience provided in their paper is a general overview ofindustrial problem-solving methods and includes 5 sectionsevent environment analysis solution and learned lessons Tomodel the experience case-based reasoning and conceptualgraphs have been used Lasierra et al [5] used ontology formodeling hospital health catalogs and sharing and acquiringknowledge from hospital personnelrsquos experiences -e role oforganized knowledge in the organized learning process hasbeen studied by Abel [6] -is knowledge is presented usingontology Ontology is the basis for knowledge maps and websystem performance for organized learning-e innovation ofthe web system provided in his paper is the organization ofresources with this knowledgemap A project has been definedthat focuses on gathering knowledge and skills in the field oforganization and more specifically on gathering resourcesrelated to this knowledge -e goal of this project is managingthis asset through the use of the information technologyoperating system that supports the organizational memory ofits users It also focuses on how users of an organization applythis system as an organizational learning vector Garcia-Crespo et al [7] designed a semantic hotel recommendationexpert system and used consumersrsquo experiences and fuzzylogic techniques to find hotels according to tourist needs

Mourtzis et al [8] Efthymiou et al [9] and Mikos et al[10] also used ontology the knowledge base and websoftware to exchange experiences Data mining and con-ceptual graphs were used by Ruiz et al [11] to reuse industryexperiences Moehrle and Raskob [12] managed nuclearevents using data mining of previous experiences Meng

et al [13] tried to extract effective information from theexperience of consumers on the consumer communitynetwork of Huawei P10P10 Plus -ey used a networkanalysis method in which each node represents a topic andthe weight of its edges denotes the number of users Renuand Mocko [14] used data mining to reuse industry expe-riences Modoni et al [15] also used ontology for knowledgereuse in industries -is study has three steps includingautomatic generation of ontologies from relational databaseschemas ontology integration and addition of rules andexecution of queries Reyes et al [16] used case-basedreasoning and the knowledge base to store and reuse ex-periences in industry Kamsu-Foguem and Abanda [17] usedthe knowledge base to gather experiences and graphicconcepts to represent ontology in industry In another studyKamsu-Foguem et al [18] used conceptual graphs andontology for checking Building Research EstablishmentEnvironmental Assessment Methodology (BREEAM) sus-tainability standard Othman and Beydoun [19] also used theknowledge base for storing and reusing disaster manage-ment knowledge Xia et al [20] proposed a new algorithmfor analyzing spatiotemporal patterns of taxi trajectories onApache Hadoop distributed computing platform in order toimprove the functionality of Hadoop in handling bigdatasets associated with massive small files and reducingmemory consumption Geng et al [21] used OWL-S topromote the geospatial service with spatiotemporal se-mantics from information to knowledge OWL-S constitutesthree logical parts including service profile (explains theusage and function of the web service) service model(service execution logic) and the service binding rules (rulesfor invoking the web service) Nowak-Brzezinska [22]proposed a new knowledge base structure and inferencealgorithm to improve analysis process for rule checking -einference algorithm is a rule clustering method whichmerges similar rules based on a similarity value Maleki et al[23] proposed ontology as a framework for supporting smartservices in product-service system of industries Accordingto studies on the storage of experiences different methodsincluding ontology [6 24] conceptual graph [25] rules [17]and logic [16] have been used for modeling experiencesHowever conceptual graphs and rules are primarily used forknowledge representation and their applications are lim-ited -is stems from the fact that all relations betweenconcepts cannot be presented by conceptual graphsMoreover modeling experiences by logical models may leadto some problems including complexity lack of explicitnessinefficiency in modeling complex phenomena such as hu-man activities behaviors and emotions and updating(changing a logical relation can lead to changing the wholemodel) However the relations defined in ontology areexplicit and everyone can understand these relations ef-fortlessly Also ontological models are easily updatable andthis rule is a fundamental principle in designing thesemodels [26]

Several studies also have been done in the field of on-tology and route finding Domain ontology has been built upfrom a high-level ontology by Camossi et al [27] -epurpose of this ontology is to identify abnormal routes in the

2 Complexity

maritime transport industry and for the detection of thetransportation of contraband in maritime surveillanceWannous et al [28] used ontology to investigate the paths ofseals Path ontology for personalized trips and animal lifemonitoring was studied by Hu et al [29] Baglioni et al [30]used path ontology to examine the behavior of individuals ina mobile game In this research a series of georeferencedpoints are embedded on the ground When a person con-nects to the game server and passes these points by an-swering some puzzles a route will be found that can be usedto examine individual behaviors during the game Duraket al [31] combined path simulation with path ontology tosimulate the angle and direction of airplane landing andflight and the flight path Malgundkar et al [32] used on-tology for urban traffic analysis Sadeghi-Niaraki et al [33]used ontology to help SDI services especially route-findinganalysis An ontology-based route-finding system was de-veloped using a multicriteria decision-making method byNiaraki and Kim [34] Saeedi et al [35] proposed an on-tology-based conceptual modeling for navigation andtourism systems -e proposed system is a multimedia andspatial database system that can be integrated with varioussensors including digital cameras and Global PositioningSystem (GPS) in mobile platforms which generalizes lo-cation-based services to context-based services to helpcontext-aware services and available mobile and desktopapplications Effati and Sadeghi-Niaraki [36] used ontologyto predict vehicle accidents Czerniak et al [37] proposed anowlANTmodel to store the graphs of possible routes of antsin an ant colony algorithm In this ontology model theclasses and the relationship of them form apexes and edgesof the graph respectively Although these studies usedontology for route finding the definition of the ontology andconceptual vision of each of these papers is different fromthe present paper and updating of the created ontology isdone less often during the process

Several researchers have used the experiences of taxidrivers in route finding For example Li et al [38] built ahierarchical street network for route finding using taxi GPSdata -ey considered only GPS data related to peak traffictime Ji-hua et al [39] used taxi GPS data to create an ex-perimental street network In this research to separateunnecessary GPS data it is assumed that the taxi speedshould not be less than the minimum speed if the taxi speedwas low it indicates that the taxi is looking for a passengerrather than transporting someone In addition a hierar-chical approach is used for route finding that firstly examineswhether the location and destination entered by the user arein the experimental network or not If both were in theexperimental street network then only the experimentalstreet network would be used for route finding If neither ofthem were in the experimental street network then theclosest point to it (or those) is found in the network and arectangle larger than the distance between the point and theclosest point in the experimental street network is selectedfrom the main street network For example if both thelocation and destination were not in the experimental streetnetwork route finding using Dijkstrarsquos algorithm is done asfollows the path between the location and the closest point

of the experimental street network is found followed by thepath between the closest point of the experimental streetnetwork of the location and the closest point of the ex-perimental street network of the destination and finally thepath between the closest point of the experimental streetnetwork of the destination and destination In our paper apart of this hierarchical approach in the proposed method isused Ziebart et al [40] proposed a probabilistic approach topredict the next intersection destination and route of thedriver based on taxi GPS data -is approach was used tohelp provide the right information and services to users atthe right moment in ubiquitous Geospatial InformationSystem (ubiquitous GIS) -e behavior of drivers is modeledin this paper and instead of providing the fastest route todirect the driver the destination of the driver is anticipatedYuan et al [41] also used GPS taxi data to create a time-dependent graph of experimental paths where a node is asegment of the route that has been repeatedly used by taxidrivers -ey used an entropy-variance-based clusteringmethod to estimate travel time In the present paper the ideaof considering the number of times that each segment ispassed by drivers for weighting edges of the street networkhas been used Chen et al [42] Chu et al [43] and Liu et al[44] identified abnormal routes and the source of theseabnormalities using taxi GPS data Liu et al [45] used ex-periences of taxi drivers to study urban structure and userplanning For example in this research busy routes aredetected and then the cause of this heavy traffic will bedetermined (for example existence of a subway station)-en it is decided that for example more residential ac-commodation will be created to reduce travel time in thisarea Rahmani and Koutsopoulos [46] formed a path graphusing taxi GPS data and used this graph to find the shortestpath with the minimum cost Zheng et al [47] used the GPSdata of 60 people within 10 months to infer transport modesof individuals based on supervised learning -is is done tohelp the cognition of human behavior and to understand theuserrsquos mobility for ubiquitous computing Due to differentstudies in this field and since reducing the processing timeof route finding is important generating a new street net-work that is simpler than the main street network to speedup route finding is frequently done However the separationof traffic-intensive (high-traffic) and low-traffic routes is notconsidered due to the departure time and the simplificationof the taxi street network and the combination of ontologyand the experiences of taxi drivers have not been used Inaddition some of these studies require the full use ofDijkstrarsquos algorithm that also reduces the speed and pre-cision of route finding in large street networks

-is paper aims to model spatial (location-based) ex-periences using ontology in the field of ubiquitous GIS routefinding Ubiquitous GIS is a combination of GIS andUbiComp and is defined as a set of concepts methods andstandards that move spatial (and temporal) data and pro-cesses into the mainstream of computing and create user-friendly programs and systems In this regard the twocomponents of any data and any user of ubiquitous GIS arefocused and each experience from any person in the field ofroute finding can be modeled in experiences ontology and

Complexity 3

can be used in route-finding algorithms-e general trend ofthis paper is that first ontology is designed for route finding-en the data needed for this ontology (paths for taxidrivers) are collected using an application Next based onthe number of times that each segment is passed by driversthe cost of each segment of the path is determined and anontology-based route finding algorithm of the driversrsquo ex-periences is proposed Finally the proposed method isimplemented in a ubiquitous GIS for the city of Tehran andis compared to Dijkstrarsquos algorithm in terms of travel timeand route length It should be noted that improving route-finding algorithm using ontology-based modeling of driverrsquospatial experiences facilitates the timeliness of servicesprovides the capability to make use of earlier informationwithout reprocessing and enhances sharing reusing andprocessing domain knowledge [48 49] -e structure of thispaper is as follows In Section 2 the concepts of ubiquitousGIS and the cause of using ontology for modeling in-dividualrsquos experiences in ubiquitous GIS are explained InSection 3 the general method of this paper driversrsquo expe-riences ontology weighting method and the proposedontology-based route-finding algorithm are introduced -eproposed algorithm is implemented using Tehranrsquos data inSection 4 In Section 5 the proposed method is compared toDijkstrarsquos algorithm in terms of travel time and route lengthFinally general conclusions of the proposed method and itsadvantages are presented

2 Ubiquitous Computing

With the growth of science and technology new generationsof computing have appeared Ubiquitous computing whichis a new generation of computing systems includes a vastrange of computers and small sensors embedded in theenvironment-e goal of ubiquitous computing is in fact todiminish the virtual dimension of technology in our lives Inubiquitous computing all the limitations are removed forhuman beings Figure 1 illustrates the elements of ubiquitouscomputing In ubiquitous computing users can access theirdesired services in any location at any time using anydevice and without any limitations in the network At thesame time the huge volume of information and services intodayrsquos world have caused the introduction of some newconcepts such as providing intelligent location-based ser-vices under any situation at any time and location using anydevice and service via any network and for any userConsidering the close relationships between GIS and in-formation technology (IT) geomatics especially GIS fol-lowed this trend and introduced ubiquitous GeospatialInformation Systems Ubiquitous GIS is based on mobilecomputing technology environment and provides mobileand distributed geographic information services It in-tegrates GIS GPS and wireless communication [50]Ubiquitous GIS enhances the provision of spatial data andservices to users even public who have no professionalknowledge of GIS in an easy-to-use way [51 52]

-e major issues with the exact route-finding algorithmsare the relatively large time it takes to find the route betweentwo locations and the high volume of data stored in a

database In a ubiquitous environment the system needs torespond to the userrsquos requests in a real-time manner hencethe noted issues are barriers that should be resolved In thispaper the ontologies are utilized to overcome this issue toenhance knowledge sharing and reusing between differentusers In other words ontologies help the stored knowledgeto be used by different users many times In this case anyuser element of the ubiquitous environment is supportedOntologies provide the capability to make use of earlierinformation without reprocessing [53] -is facilitates thetimeliness of services In addition using ontology in route-finding algorithm was to store the knowledge of experiencedpersons -is makes the systems smart ubiquitous and actlike an experienced human Moreover the data in ontologiesare stored in plain text formats such as Web OntologyLanguage (OWL) Obviously plain text formats requiremuch lower storage space Moreover ontologies are in-dependent from text and implementation and operate on ahigher level of abstraction However databases are situatedat the lower level of abstraction and they are mainly designedto meet the requirements of a particular application orcorporation and when the requirements change the schemaof a database also need to be modified [54 55] In our re-search since we update the ontology many times relationaldatabases cannot be used Relational databases are not ap-propriate for applications that need to make frequent up-dates since you need to change the schema for each updateIn addition ontologies do not need to be normalized (suchcharacteristics make for easier the information sharing andmerging) [56] However relational databases need to benormalized Besides ontology languages are more expressivein terms of expressing more semantic concepts than data-base languages which only include constructs for defining orextracting data [55] In addition ontology languages providea more correct and precise domain conceptualization [57]

In this research any user can access the ontology at anytime and in any location to find routes In addition due tothe intelligence of ubiquitous environment the experiencesfrom users (any data element of ubiquitous computing) canbe received and stored to be utilized for future services

3 Methodology

Figure 2 shows the main processes of this research Based onthese processes the ontology-based route-finding algorithmis proposed which enables to store and retrieve spatial ex-periences of the taxi drivers

An ontology-based algorithm for storing spatial expe-riences in ubiquitous GIS space has been presented Figure 3depicts the workflow of the proposed approach First thedataset is collected via an app -is app collects the routes oftaxi vehicles needed for the ontology -e number of timesthat each segment is passed by drivers is calculated and usedto weight the edges of the street network and to build the costmodel -en taxi paths are converted to the driversrsquo ex-perience class in the ontology Finally an ontology-basedroute-finding algorithm is designed implemented andevaluated in ubiquitous environment -e work has beendescribed in detail in the following sections

4 Complexity

-e algorithm makes the route-finding systems smartand ubiquitous intended to act like experienced people Aubiquitous system is a smart system which acts not only likea person but also like an experienced human A person whoworks for some field of interest for many years gets a lot ofexperience-is person has much knowledge To transfer theknowledge of the experienced persons to the system we needthis algorithm-e proposed algorithm can receive and storethe knowledge of an experienced person and also accu-mulate the experiences of many experienced persons

31 Driversrsquo Experience Ontology In our ontology thedriversrsquo experience class includes two subclasses experi-mental routes and nonexperimental routes -e experi-mental route class encompasses the paths which the

experienced drivers have driven -ese routes are collectedvia an app and are converted to OWL classes -is classitself includes two subclasses traffic and without-trafficclasses -e paths which have been passed at peak traffictimes (7ndash10 AM and 16ndash20 PM) are categorized under thetraffic subclass -e nonexperimental route class encom-passes the paths obtained from the route-finding procedure-ese should also be converted to OWL classes In thedriversrsquo experience ontology for the sake of simplicity and inorder to improve usability only the names of routes aredefined as classes individuals of each class are not defined inthis file For each route class a separate OWL file is definedthe name of which is the same as the corresponding routeclass in the driversrsquo experience ontology -e constituentnodes of a route are stored as individuals in the corre-sponding OWL file using their identification (ID) code as

Any data Any place

Any time

Any network

Any service

Any device

Any user

Ubiquitousinformation

system

Others

Figure 1 Elements of ubiquitous computing

Data collection

Data preparation

Data modeling

Evaluation

Implementation

Monitoring

Ontology designcreating road networks

Predictiveanalytics

Determiningweights

Compare to Google maps based on route length and travel time

Adding new routesto ontology

Figure 2 Research methodology

Complexity 5

their name -e coordinates of each node are defined as dataproperties of individuals in the ontology

-e created ontology is shown in Figure 4 To save spaceonly a limited number of routes are shown here -is figuredepicts the ontology before the route-finding process Afterroute finding if a new path has been created and the timeentered by the user is within the pick range it will be storedin the traffic subclass of the nonexperimental route classOtherwise it will be stored in the nontraffic subclass Eachclass in the lowest level of ontology shows a route and itsname is defined based on its location and destination Forroutes which were traversed on nontraffic times the name ofclass would be location destination and for routes related totraffic classes the name of class would be location_desti-nation_t If the route was traversed on traffic times and it wasretrieved from algorithmrsquos results the name of class wouldbe location_destination_n_t -e flowchart of this ontologyis shown in Figure 5

32 Weighing Method In this paper a graph for a streetnetwork is constructed -e route finding is performed

based on this graph Each segment of the path is consideredas an edge of the graph -e weight of each edge is de-termined based on the number of times the drivers havepassed the corresponding segment Since this number mightbe unexpectedly great its normal form is computed asfollows

fn ei( 1113857 f ei( 1113857 minus fmin + 1fmax minus fmin + 1

(1)

where f(ei) is the number of passes of edge ei and fmin andfmax are the minimum and maximum number of the wholepasses in the graph For a segment which is not located inexperimental routes f(ei) 0 -erefore fmin 0 Gener-ally the weights of the edges of the street network arecomputed using the following equation

w ei( 1113857 maxfn(e)

Length (e)1113888 1113889 minus

fn ei( 1113857

Length ei( 1113857 (2)

where w(ei) denotes the weight of ith edge of the graphLength(ei) is the length of the edge and max(fn

(e)Length(e)) is the maximum number of passes of an edge

Evaluation

Calculating route length and travel time

Combining ontology with the cost model and

applying it to road network

Implementation ofroute-finding algorithm

Data collection

Computing frequency oftraversing a segment Data preprocessing

Determiningweighing method

Converting routes toOWL files

Cost model creation Ontology creation

Figure 3 Workflow of the proposed approach

6 Complexity

divided by the length of that edge Since edges with smallerweights are preferred in Dijkstrarsquos algorithm the values aresubtracted from the maximum value to inverse values

-e reason for choosing Dijkstrarsquos algorithm in thisresearch is that in the proposed algorithm in each exe-cution if a new path is created by an algorithm it will bestored in ontology in order to reuse it in future -ereforewe need to use an algorithm which provides an exact resultwith one hundred percent reliability because the results willbe reused in the future Among the shortest path algo-rithms breadth-first search (BFS) and depth-first search(DFS) can only be used in a weighted graph if the weightsare equal However in our graph weights which are thenumber of traverses by taxis are not equal Moreover theyare slower than in the Dijkstra algorithm [58] -e Greedyalgorithm uses heuristics and Alowast algorithms which are acombination of the Greedy algorithm and the Dijkstraalgorithm -erefore they provide an approximate resultnot an exact result

33 Ubiquitous Ontology-Based Route-Finding AlgorithmFigure 6 depicts the way our proposed algorithm worksFirst the user specifies the location the destination andthe time -en the algorithm determines whether asubclass with the specified location and destination existsin the ontology or not If the time chosen falls within therange of peak times the existing paths in the trafficsubclass are searched Otherwise the paths in the non-traffic subclass are examined When a path exists thestored route is displayed to the user and the algorithmterminates without needing to conduct route findingHowever when there is no corresponding path the al-gorithm determines whether the location and destinationexist in the experimental street network or not If bothexist in the network only the experimental network is usedto construct the graph and find the route (again peak timedetermines the subclass to choosemdashtraffic or nontraffic)

and the algorithm terminates If the location andor thedestination does not exist in the experimental streetnetwork the closest points of the network to them arefound -en a rectangle with a diameter (R2 R1 + 2radic2 h)greater than the distance between the location (or desti-nation) and the point is extracted from the main roadnetwork -e flowchart of the proposed hierarchical al-gorithm and the extraction of the rectangular region fromthe road network are shown in Figure 7 To clarify theroute-finding procedure consider a situation where boththe location and the destination do not exist in the ex-perimental network In such a situation the route findingincludes three stages (1) route finding from the location toits closest point in the experimental network (SSrsquo) (2)route finding from this point to the closest point of theexperimental network with respect to the destination(SrsquoDrsquo) and (3) route finding from the closest point of theexperimental network (with respect to the destination) tothe destination (DrsquoD) After route finding the resultingroute is stored in the driversrsquo experience ontology as aseparate OWL file Now requests with these locations anddestinations are responded to in real time without re-quiring additional route finding

4 Implementation

In this paper an ontology-based route-finding algorithmbased on the driversrsquo experience in ubiquitous GIS spacehas been proposed -e steps of implementing the algo-rithm are shown in Figure 8 In order to implement theproposed algorithm the required data have been collectedvia an app that stores taxi vehiclesrsquo routes throughOpenStreetMap (OSM) maps -ese routes and theircorresponding pick-up and drop-off stations are located inTehran Iran and are entered into the app by the drivers-en they are converted to shapefiles Preprocessing tasksare performed in the ArcGIS software package and theresults are stored in the Oracle database Each route is also

Figure 4 Driversrsquo experience ontology (to save space only a limited number of routes are shown here)

Complexity 7

converted to an OWL file using OWL Application Pro-gramming Interface (API) in Java Using these routes adriversrsquo experience ontology is separately constructed inProtege software A sample of a route stored in the app isshown in Figure 9

After data preparation the presented algorithm wasimplemented In each run of the app if a new route has beengenerated it will be stored as a class in the driversrsquo expe-rience ontology to be used in future requests without needfor additional processing As a result the routes are stored ina low-volume manner while data processing reduces witheven more runs of the algorithm Each route class in theontology occupies about 20ndash25 percent of its correspondingshapefile Moreover in the presented algorithm two roadnetworks existmdasha peak-time road network and a low-traffic

road network During peak times finding an optimum routebecomes important therefore the algorithm will utilizeDijkstrarsquos algorithm and present the optimum path to theuser

5 Evaluation

In order to evaluate the proposed method 10 different lo-cation-destination pairs covering most of Tehran city wereselected First the experimental road network was fullyconstructed (ER) -en it was constructed only consideringthe low-traffic experimental routes (ETR) Also the shortestpaths between the location-destinations were obtained usingDijkstrarsquos algorithm (DR) For all obtained routes the routelengths were calculated Using driversrsquo reports the travel

Collecting drivers routeswith app

Adding the name (location_destination) of

each new route as a subclass of nontraffic class of experimental

routes

Consider the segments of each path as individuals of

that classrsquo path

Considering the fields (X coordinates Y coordinates)

as the data property

Driversrsquo experienceontology

Experimental route classNonexperimental route class

Adding the name (location_destination) of

each new route as a subclass of nontraffic

class ofnonexperimental routes

Creating experimentaland nonexperimental

route classes

Storing each path in a separate OWL file as a class

Retrieving each class ifrequested by the user

Traffic class of experimental routes

Nontraffic class of experimental routes

Nontraffic class ofnonexperimental routes

Traffic class ofnonexperimental routes

Adding the name (location_destination_t)of each new route as a subclass of traffic class of experimental routes

Is the enteredtime by user is in range

of pick traffic times

Adding the name (location_destination_t)of each new route as a subclass of traffic class

of nonexperimental routes

No Yes

Figure 5 Flowchart of creating driversrsquo experience ontology

8 Complexity

times were also calculated Figure 10 illustrates the resultingroutes for a location-destination pair (Route 10) using thethree methods above

51 Evaluation Based on the Route Length Figure 11 shows adiagram of the length of the paths for the three methods-eroute lengths have been computed by accumulating their

Get the locationdestination and time

from user

Is anyroute with this locationand destination in the

nontraffic classes of driverrsquosexperience ontology

Showing the saved pathto the user

End

Are locationand destinationin experimentalroadnetwork

Using only experimentalroad network for route

finding

Using Dijkstrarsquosalgorithm for route

finding

Finding the nearest point of theexperimental road network to the location

and extraction of a rectangle from themain road network with diameter largerthan the distance between the location

and the nearest point in the experimentalroad network

Is location in experimental

road network

Is destination inexperimental

road network

Route finding with experimentalroad network and segments inthe rectangle(s) of main road

network

Storing route name as a subclass oftraffic class of nonexperimental

road network in driverrsquosexperiences ontology

Finding the nearest point of theexperimental road network to the

destination and extraction of a rectanglefrom the main road network with

diameter larger than the distance betweenthe destination and the nearest point in

the experimental road network

Storing route in aseparate OWL file with

its segments andproperties

Yes

Yes

No

NoNoYes

Is the entered

time in peaktraffic times

Is anyroute with this locationand destination in the

traffic classes of driverrsquosexperience ontology

Considering experimental roadnetwork according to trafficclass of experimental routes

Considering experimental roadnetwork according to nontraffic

class of experimental routes

No Yes

NoNo

Is entered time inpeak traffic times

Storing route name as a subclass ofnontraffic class of nonexperimental

road network indriverrsquos experiences ontology

Yes No

Yes

No

Figure 6 Proposed algorithm flowchart

Complexity 9

constituting segments Figure 12 depicts the ratio of routelengths obtained from the two proposed methods to that ofDijkstrarsquos algorithm Figure 13 compares the mean length ofthe 10 routes for each method

It is observable from the figures that route lengths in theshortest path algorithm are shorter than that of the other twomethods -ey have in contrast greater values when con-sidering high-traffic routes -e mean length of routes is1102 km for Dijkstrarsquos algorithm 1422 km when only con-sidering traffic routes and 134 km when considering all theexperimental (traffic and nontraffic) routes -is is due to taxidrivers who generally select the longest but simultaneously

fastest routes On the other hand Dijkstrarsquos algorithm doesnot consider travel time

52 Evaluation Based on Travel Time Figure 14 shows adiagram of the travel times for the three methods -etravel times have been obtained from different driverswho have driven these routes in peak times and in realconditions Figure 15 depicts the travel times obtainedfrom the two proposed methods to that of Dijkstrarsquos al-gorithm Figure 16 compares the mean travel times foreach method

Experimental road network layer

Basic road network layer

Drsquo

Drsquo

D

Srsquo

SrsquoS

(a)

S

R1

R2

h

D

(b)

Figure 7 (a) Hierarchical experimental road network (b) Extraction of rectangular region from main road network [39]

Trace preprocessing

Ontology design

Ontology-based route finding

Ubiquitous ontology-based route finding

User interface Storingroute if anew routeis created

OWL route filesUser Ontologicalroute-finding

algorithm

GIS and ontologyexperts

Driversrsquo experiences ontology OWL route files Ontology data

Taxiroutedata

Mobile appfor data

collection

Experiencedtaxi drivers

Examiningfrequency ofusing a road

segment

Costmodel

a7

149

10

2 116

15

b

c d

ef

Routenetwork

graph

Figure 8 Steps for implementing the proposed algorithm

10 Complexity

Figure 9 A sample of a stored route in the app which collects data

DR

ETR

ERLocationdestination

Final map

N

S

W E

Figure 10 Resulting routes for a location-destination pair (Route 10) using the three methods

DijkstraExperimental routesExperimental traffic routes

2 3 4 5 6 7 8 9 101Route number

02468

101214161820

Rout

e len

gth

(km

)

Figure 11 Comparison of route length for 10 evaluated routes of the three methods

Complexity 11

It is observable from the figures that travel timesobtained from the hierarchical experimental methodwhen only considering high-traffic conditions are lowesthowever the travel time of the shortest path is the highest-e mean travel time is 80 minutes for Dijkstrarsquos algo-rithm 61 minutes when only considering the high-trafficroutes and 775 minutes when considering all of theexperimental routes Although the routes are long whenconsidering high-traffic routes they are fast due to theselection of low-traffic routes in this case For examplewhen going to Tajrish Square from Valiasr SquareDijkstrarsquos algorithm suggests driving through ValiasrStreet which corresponds to the shortest but not fastestroute However the proposed ontology-based approachsuggests that the driver proceeds mainly along ModarresHighway which reduces the travel time by 35 minutes inhigh-traffic conditions

Another approach to evaluate the two proposedmethodsagainst Dijkstrarsquos algorithm is to find the ideal route amongthem In order to find the ideal route one can divide theroute length by its travel time Obviously when the routelengths for all routes or just two of them are equal the onewhich has the shortest travel time should be selected as theoptimum path On the other hand when the travel times areequal the one which has the shortest length should be se-lected as the optimum path Accordingly in eight cases (of10 cases) the ontology-based route finding considering onlyhigh-traffic routes (ETR) predominates In the other twocases the travel time was roughly equal however the routelengths obtained fromDijkstrarsquos algorithm had lower valuesTo determine the deviation of each method from the idealcase which is shown in Figure 17 the following formula isused

Fj travel length of method j

travel time of method j

j ETR ER Dikjestra

Fe ideal route fraction

Dj Fj minus Fe1113872 1113873

Fe

(3)

-erefore according to the evaluations above it can beclaimed that eliminating the experimental high-traffic routesin high-traffic conditions and utilizing the driversrsquo

02468

10121416

Mea

n of

rout

e len

gth

(km

)

Dijkstra Experimentalroutes

Method name

Experimentaltrafficroutes

Figure 13 Mean of route length for 3 methods

0

20

40

60

80

100

120

Trav

el ti

me (

min

)

1 2 3 4 5 6 7 8 9 10Route number

DijkstraExperimental routesExperimental traffic routes

Figure 14 Comparison of travel time for 10 evaluated routes of 3methods

002040608

112

1 2 3 4 5 6 7 8 9 10

ERDijkstraETRDijkstra

Figure 15 Comparison of travel time for the 2 modes of theproposed method with Dijkstrarsquos algorithm

Dijkstra020406080

Mea

n of

trav

el ti

me (

min

)

Experimentalroutes

Method name

Experimentaltrafficroutes

Figure 16 Mean of travel time for 3 methods

0

05

1

15

2

1 2 3 4 5 6 7 8 9 10

ERDijkstraETRDijkstra

Figure 12 Comparison of route length in 2 modes of the proposedalgorithm with Dijkstrarsquos algorithm

12 Complexity

experiences make the algorithm generate the optimum routethrough low-traffic parts of the network -is minimizestravel time which is the most important parameter in routefinding

6 Conclusions

Different people gain different spatial experiences Asthere are no systems to store these experiences theyevaporate over time-e present study provides the abilityto store spatial experiences and to make the existingsystems in ubiquitous GIS space intelligent In this paperan ontology-based route-finding algorithm in ubiquitousGIS space was designed and implemented using theroutes driven by the drivers of Tehran in high- and low-traffic conditions

In the proposed route-finding algorithm based on thedriversrsquo experiences ontology the number of times that eachsegment is passed by drivers is used to assign weights to thatsegment of the network First the user specifies the location thedestination and the time -en the algorithm determineswhether a subclass with the specified location and destinationexists in the ontology or not If the time chosen falls duringpeak times the existing paths in the traffic subclass are searchedthrough Otherwise the paths in the nontraffic subclass areexamined When a path exists the stored route is displayed tothe user and the algorithm terminates without needing toconduct route finding However when there is no corre-sponding path the algorithm determines whether the locationand destination exist in the experimental road network or notIf both exist in the network only the experimental is used toconstruct the graph and find the route (again peak time

determines the subclass to choosemdashtraffic or nontraffic) andthe algorithm terminates If the origin andor the destinationdoes not exist in the experimental road network the closestpoints of the network to them are found-en a rectangle witha diameter (R2R1 + 2radic2h) greater than the distance betweenthe location (or destination) and the point is extracted from themain road network -e main difference of this algorithm tothe previous experimental route-finding algorithms is the useof ontology and the construction of road networks onlyconsidering the low-traffic paths when there are high-trafficpaths in the network Since each newpath is stored as a separateOWL class the storage issues in the databases are resolved Inaddition the use of ontology causes the system to update everytime the algorithm runs and generates new paths Utilizing onlylow-traffic routes in high-traffic conditions allows the system tosuggest routes that are faster ensuring the optimum path basedon travel time is achieved According to our evaluations theproposed algorithm suggests routes that are longer but si-multaneously the fastest -is guarantees the appropriatenessof using such an algorithm in route finding In this researchany user can access the ontology from any location and at anytime to conduct route finding Moreover being intelligent theubiquitous environment can receive and store the experiences(any data element of ubiquitous GIS) of any person to utilizewithin its services in the future

More innovative applications of spatial experiencemodeling such as ambulance firefighting and tourismservices can be subjects for further considerations -eprocess of route finding for this category is also affected bytraffic as emergency services can also become stuck in trafficlike other drivers because there are no dedicated lines on allroads Due to the limitation in length of road and a largenumber of vehicles in the road in traffic times it is hard forother vehicles to provide a route for traversing of the am-bulance and they still need to find a way to avoid traffic-erefore the experiences from expert technicians in am-bulances or firefighters who have been in real event con-ditions can be modeled and used in the future In additionthis study considers only two factors distance and time forevaluation Other criteria including fuel consumption can beconsidered in future

Data Availability

-e authors are ready to share the research data on request

Disclosure

Maryam Barzegar and Abolghasem Sadeghi-Niaraki are tobe considered as co-first authors

Conflicts of Interest

-e authors declare that they have no conflicts of interest

Acknowledgments

-is research was supported by theMSIT (Ministry of Scienceand ICT) Korea under the ITRC (Information Technology

0

01

02

03

041

2

3

4

5

6

7

8

9

10

DijkstraExperimental routes

Experimental traffic routesIdeal

Figure 17 Comparison of routes from the 3methods with the idealroute

Complexity 13

Research Center) support program (IITP-2019-2016-0-00312)supervised by the IITP (Institute for Information andCommunications Technology Planning and Evaluation)

References

[1] I P Tussyadiah and F J Zach ldquo-e role of geo-basedtechnology in place experiencesrdquo Annals of Tourism Researchvol 39 no 2 pp 780ndash800 2012

[2] G M Honti and J Abonyi ldquoA review of semantic sensortechnologies in internet of things architecturesrdquo Complexityvol 2019 Article ID 6473160 21 pages 2019

[3] U H Govindarajan A J Trappey and C V TrappeyldquoImmersive technology for human-centric cyberphysicalsystems in complex manufacturing processes a compre-hensive overview of the global patent profile using collectiveintelligencerdquo Complexity vol 2018 Article ID 428363417 pages 2018

[4] B K Foguem T Coudert C Beler and L GenesteldquoKnowledge formalization in experience feedback processesan ontology-based approachrdquo Computers in Industry vol 59no 7 pp 694ndash710 2008

[5] N Lasierra F Roldan A Alesanco and J Garcıa ldquoTowardsimproving usage and management of supplies in healthcarean ontology-based solution for sharing knowledgerdquo ExpertSystems with Applications vol 41 no 14 pp 6261ndash6273 2014

[6] M-H Abel ldquoKnowledge map-based web platform to facilitateorganizational learning return of experiencesrdquo Computers inHuman Behavior vol 51 pp 960ndash966 2015

[7] A Garcıa-Crespo J L Lopez-Cuadrado R Colomo-PalaciosI Gonzalez-Carrasco and B Ruiz-Mezcua ldquoSem-fit a se-mantic based expert system to provide recommendations inthe tourism domainrdquo Expert Systems with Applicationsvol 38 no 10 pp 13310ndash13319 2011

[8] D Mourtzis M Doukas and C Giannoulis ldquoAn inference-based knowledge reuse framework for historical product andproduction information retrievalrdquo Procedia CIRP vol 41pp 472ndash477 2016

[9] K Efthymiou K Sipsas D Mourtzis and G ChryssolourisldquoOn knowledge reuse for manufacturing systems design andplanning a semantic technology approachrdquo CIRP Journal ofManufacturing Science and Technology vol 8 pp 1ndash11 2015

[10] W L Mikos J C E Ferreira P E A Botura and L S FreitasldquoA system for distributed sharing and reuse of design andmanufacturing knowledge in the PFMEA domain using adescription logics-based ontologyrdquo Journal of ManufacturingSystems vol 30 no 3 pp 133ndash143 2011

[11] P P Ruiz B K Foguem and B Grabot ldquoGeneratingknowledge in maintenance from experience feedbackrdquoKnowledge-Based Systems vol 68 pp 4ndash20 2014

[12] S Moehrle and W Raskob ldquoStructuring and reusingknowledge from historical events for supporting nuclearemergency and remediation managementrdquo Engineering Ap-plications of Artificial Intelligence vol 46 pp 303ndash311 2015

[13] Q Meng Z Zhang X Wan and X Rong ldquoProperties ex-ploring and information mining in consumer communitynetwork a case of huawei pollen clubrdquo Complexity vol 2018Article ID 9470580 19 pages 2018

[14] R S Renu and GMocko ldquoComputing similarity of text-basedassembly processes for knowledge retrieval and reuserdquoJournal of Manufacturing Systems vol 39 pp 101ndash110 2016

[15] G E Modoni M Doukas W Terkaj M Sacco andD Mourtzis ldquoEnhancing factory data integration through thedevelopment of an ontology from the reference models reuse

to the semantic conversion of the legacy modelsrdquo In-ternational Journal of Computer Integrated Manufacturingvol 30 no 10 pp 1043ndash1059 2017

[16] E R Reyes S Negny G C Robles and J M Le LannldquoImprovement of online adaptation knowledge acquisitionand reuse in case-based reasoning application to processengineering designrdquo Engineering Applications of ArtificialIntelligence vol 41 pp 1ndash16 2015

[17] B Kamsu-Foguem and F H Abanda ldquoExperience modelingwith graphs encoded knowledge for construction industryrdquoComputers in Industry vol 70 pp 79ndash88 2015

[18] B Kamsu-Foguem F H Abanda M B Doumbouya andJ F Tchouanguem ldquoGraph-based ontology reasoning forformal verification of BREEAM rulesrdquo Cognitive SystemsResearch vol 55 pp 14ndash33 2019

[19] S H Othman and G Beydoun ldquoA metamodel-basedknowledge sharing system for disaster managementrdquo ExpertSystems with Applications vol 63 pp 49ndash65 2016

[20] D Xia X Lu H Li W Wang Y Li and Z Zhang ldquoAmapreduce-based parallel frequent pattern growth algorithmfor spatiotemporal association analysis of mobile trajectorybig datardquo Complexity vol 2018 Article ID 2818251 16 pages2018

[21] J Geng S Wang W Gan et al ldquoPromoting geospatial servicefrom information to knowledge with spatiotemporal se-manticsrdquo Complexity vol 2019 Article ID 9301420 14 pages2019

[22] A Nowak-Brzezinska ldquoEnhancing the efficiency of a decisionsupport system through the clustering of complex rule-basedknowledge bases and modification of the inference algo-rithmrdquo Complexity vol 2018 Article ID 2065491 14 pages2018

[23] E Maleki F Belkadi N Boli et al ldquoOntology-basedframework enabling smart product-service systems appli-cation of sensing systems for machine health monitoringrdquoIEEE Internet of ings Journal vol 5 no 6 pp 4496ndash45052018

[24] F H Abanda B Kamsu-Foguem and J H M TahldquoBIMmdashnew rules of measurement ontology for constructioncost estimationrdquo Engineering Science and Technology anInternational Journal vol 20 no 2 pp 443ndash459 2017

[25] B Kamsu-Foguem and P Tiako ldquoRisk information formal-isation with graphsrdquo Computers in Industry vol 85 pp 58ndash69 2017

[26] M Barzegar A Sadeghi-Niaraki and M Shakeri ldquoSpatialexperience based route finding using ontologiesrdquo ETRIJournal pp 1ndash11 2019

[27] E Camossi P Villa and L Mazzola ldquoSemantic-basedanomalous pattern discovery in moving object trajectoriesrdquo2013 httpsarxivorgabs13051946

[28] R Wannous J Malki A Bouju and C Vincent ldquoModellingmobile object activities based on trajectory ontology rulesconsidering spatial relationship rulesrdquo in Modeling Ap-proaches and Algorithms for Advanced Computer Applicationspp 249ndash258 Springer International Publishing BerlinGermany 2013

[29] Y Hu K Janowicz D Carral et al ldquoA geo-ontology designpattern for semantic trajectoriesrdquo in Proceedings of the In-ternational Conference on Spatial Information eorypp 438ndash456 Springer International Publishing ScarboroughUK September 2013

[30] M Baglioni J Macedo C Renso and M Wachowicz ldquoAnontology-based approach for the semantic modelling andreasoning on trajectoriesrdquo in Advances in Conceptual

14 Complexity

maritime transport industry and for the detection of thetransportation of contraband in maritime surveillanceWannous et al [28] used ontology to investigate the paths ofseals Path ontology for personalized trips and animal lifemonitoring was studied by Hu et al [29] Baglioni et al [30]used path ontology to examine the behavior of individuals ina mobile game In this research a series of georeferencedpoints are embedded on the ground When a person con-nects to the game server and passes these points by an-swering some puzzles a route will be found that can be usedto examine individual behaviors during the game Duraket al [31] combined path simulation with path ontology tosimulate the angle and direction of airplane landing andflight and the flight path Malgundkar et al [32] used on-tology for urban traffic analysis Sadeghi-Niaraki et al [33]used ontology to help SDI services especially route-findinganalysis An ontology-based route-finding system was de-veloped using a multicriteria decision-making method byNiaraki and Kim [34] Saeedi et al [35] proposed an on-tology-based conceptual modeling for navigation andtourism systems -e proposed system is a multimedia andspatial database system that can be integrated with varioussensors including digital cameras and Global PositioningSystem (GPS) in mobile platforms which generalizes lo-cation-based services to context-based services to helpcontext-aware services and available mobile and desktopapplications Effati and Sadeghi-Niaraki [36] used ontologyto predict vehicle accidents Czerniak et al [37] proposed anowlANTmodel to store the graphs of possible routes of antsin an ant colony algorithm In this ontology model theclasses and the relationship of them form apexes and edgesof the graph respectively Although these studies usedontology for route finding the definition of the ontology andconceptual vision of each of these papers is different fromthe present paper and updating of the created ontology isdone less often during the process

Several researchers have used the experiences of taxidrivers in route finding For example Li et al [38] built ahierarchical street network for route finding using taxi GPSdata -ey considered only GPS data related to peak traffictime Ji-hua et al [39] used taxi GPS data to create an ex-perimental street network In this research to separateunnecessary GPS data it is assumed that the taxi speedshould not be less than the minimum speed if the taxi speedwas low it indicates that the taxi is looking for a passengerrather than transporting someone In addition a hierar-chical approach is used for route finding that firstly examineswhether the location and destination entered by the user arein the experimental network or not If both were in theexperimental street network then only the experimentalstreet network would be used for route finding If neither ofthem were in the experimental street network then theclosest point to it (or those) is found in the network and arectangle larger than the distance between the point and theclosest point in the experimental street network is selectedfrom the main street network For example if both thelocation and destination were not in the experimental streetnetwork route finding using Dijkstrarsquos algorithm is done asfollows the path between the location and the closest point

of the experimental street network is found followed by thepath between the closest point of the experimental streetnetwork of the location and the closest point of the ex-perimental street network of the destination and finally thepath between the closest point of the experimental streetnetwork of the destination and destination In our paper apart of this hierarchical approach in the proposed method isused Ziebart et al [40] proposed a probabilistic approach topredict the next intersection destination and route of thedriver based on taxi GPS data -is approach was used tohelp provide the right information and services to users atthe right moment in ubiquitous Geospatial InformationSystem (ubiquitous GIS) -e behavior of drivers is modeledin this paper and instead of providing the fastest route todirect the driver the destination of the driver is anticipatedYuan et al [41] also used GPS taxi data to create a time-dependent graph of experimental paths where a node is asegment of the route that has been repeatedly used by taxidrivers -ey used an entropy-variance-based clusteringmethod to estimate travel time In the present paper the ideaof considering the number of times that each segment ispassed by drivers for weighting edges of the street networkhas been used Chen et al [42] Chu et al [43] and Liu et al[44] identified abnormal routes and the source of theseabnormalities using taxi GPS data Liu et al [45] used ex-periences of taxi drivers to study urban structure and userplanning For example in this research busy routes aredetected and then the cause of this heavy traffic will bedetermined (for example existence of a subway station)-en it is decided that for example more residential ac-commodation will be created to reduce travel time in thisarea Rahmani and Koutsopoulos [46] formed a path graphusing taxi GPS data and used this graph to find the shortestpath with the minimum cost Zheng et al [47] used the GPSdata of 60 people within 10 months to infer transport modesof individuals based on supervised learning -is is done tohelp the cognition of human behavior and to understand theuserrsquos mobility for ubiquitous computing Due to differentstudies in this field and since reducing the processing timeof route finding is important generating a new street net-work that is simpler than the main street network to speedup route finding is frequently done However the separationof traffic-intensive (high-traffic) and low-traffic routes is notconsidered due to the departure time and the simplificationof the taxi street network and the combination of ontologyand the experiences of taxi drivers have not been used Inaddition some of these studies require the full use ofDijkstrarsquos algorithm that also reduces the speed and pre-cision of route finding in large street networks

-is paper aims to model spatial (location-based) ex-periences using ontology in the field of ubiquitous GIS routefinding Ubiquitous GIS is a combination of GIS andUbiComp and is defined as a set of concepts methods andstandards that move spatial (and temporal) data and pro-cesses into the mainstream of computing and create user-friendly programs and systems In this regard the twocomponents of any data and any user of ubiquitous GIS arefocused and each experience from any person in the field ofroute finding can be modeled in experiences ontology and

Complexity 3

can be used in route-finding algorithms-e general trend ofthis paper is that first ontology is designed for route finding-en the data needed for this ontology (paths for taxidrivers) are collected using an application Next based onthe number of times that each segment is passed by driversthe cost of each segment of the path is determined and anontology-based route finding algorithm of the driversrsquo ex-periences is proposed Finally the proposed method isimplemented in a ubiquitous GIS for the city of Tehran andis compared to Dijkstrarsquos algorithm in terms of travel timeand route length It should be noted that improving route-finding algorithm using ontology-based modeling of driverrsquospatial experiences facilitates the timeliness of servicesprovides the capability to make use of earlier informationwithout reprocessing and enhances sharing reusing andprocessing domain knowledge [48 49] -e structure of thispaper is as follows In Section 2 the concepts of ubiquitousGIS and the cause of using ontology for modeling in-dividualrsquos experiences in ubiquitous GIS are explained InSection 3 the general method of this paper driversrsquo expe-riences ontology weighting method and the proposedontology-based route-finding algorithm are introduced -eproposed algorithm is implemented using Tehranrsquos data inSection 4 In Section 5 the proposed method is compared toDijkstrarsquos algorithm in terms of travel time and route lengthFinally general conclusions of the proposed method and itsadvantages are presented

2 Ubiquitous Computing

With the growth of science and technology new generationsof computing have appeared Ubiquitous computing whichis a new generation of computing systems includes a vastrange of computers and small sensors embedded in theenvironment-e goal of ubiquitous computing is in fact todiminish the virtual dimension of technology in our lives Inubiquitous computing all the limitations are removed forhuman beings Figure 1 illustrates the elements of ubiquitouscomputing In ubiquitous computing users can access theirdesired services in any location at any time using anydevice and without any limitations in the network At thesame time the huge volume of information and services intodayrsquos world have caused the introduction of some newconcepts such as providing intelligent location-based ser-vices under any situation at any time and location using anydevice and service via any network and for any userConsidering the close relationships between GIS and in-formation technology (IT) geomatics especially GIS fol-lowed this trend and introduced ubiquitous GeospatialInformation Systems Ubiquitous GIS is based on mobilecomputing technology environment and provides mobileand distributed geographic information services It in-tegrates GIS GPS and wireless communication [50]Ubiquitous GIS enhances the provision of spatial data andservices to users even public who have no professionalknowledge of GIS in an easy-to-use way [51 52]

-e major issues with the exact route-finding algorithmsare the relatively large time it takes to find the route betweentwo locations and the high volume of data stored in a

database In a ubiquitous environment the system needs torespond to the userrsquos requests in a real-time manner hencethe noted issues are barriers that should be resolved In thispaper the ontologies are utilized to overcome this issue toenhance knowledge sharing and reusing between differentusers In other words ontologies help the stored knowledgeto be used by different users many times In this case anyuser element of the ubiquitous environment is supportedOntologies provide the capability to make use of earlierinformation without reprocessing [53] -is facilitates thetimeliness of services In addition using ontology in route-finding algorithm was to store the knowledge of experiencedpersons -is makes the systems smart ubiquitous and actlike an experienced human Moreover the data in ontologiesare stored in plain text formats such as Web OntologyLanguage (OWL) Obviously plain text formats requiremuch lower storage space Moreover ontologies are in-dependent from text and implementation and operate on ahigher level of abstraction However databases are situatedat the lower level of abstraction and they are mainly designedto meet the requirements of a particular application orcorporation and when the requirements change the schemaof a database also need to be modified [54 55] In our re-search since we update the ontology many times relationaldatabases cannot be used Relational databases are not ap-propriate for applications that need to make frequent up-dates since you need to change the schema for each updateIn addition ontologies do not need to be normalized (suchcharacteristics make for easier the information sharing andmerging) [56] However relational databases need to benormalized Besides ontology languages are more expressivein terms of expressing more semantic concepts than data-base languages which only include constructs for defining orextracting data [55] In addition ontology languages providea more correct and precise domain conceptualization [57]

In this research any user can access the ontology at anytime and in any location to find routes In addition due tothe intelligence of ubiquitous environment the experiencesfrom users (any data element of ubiquitous computing) canbe received and stored to be utilized for future services

3 Methodology

Figure 2 shows the main processes of this research Based onthese processes the ontology-based route-finding algorithmis proposed which enables to store and retrieve spatial ex-periences of the taxi drivers

An ontology-based algorithm for storing spatial expe-riences in ubiquitous GIS space has been presented Figure 3depicts the workflow of the proposed approach First thedataset is collected via an app -is app collects the routes oftaxi vehicles needed for the ontology -e number of timesthat each segment is passed by drivers is calculated and usedto weight the edges of the street network and to build the costmodel -en taxi paths are converted to the driversrsquo ex-perience class in the ontology Finally an ontology-basedroute-finding algorithm is designed implemented andevaluated in ubiquitous environment -e work has beendescribed in detail in the following sections

4 Complexity

-e algorithm makes the route-finding systems smartand ubiquitous intended to act like experienced people Aubiquitous system is a smart system which acts not only likea person but also like an experienced human A person whoworks for some field of interest for many years gets a lot ofexperience-is person has much knowledge To transfer theknowledge of the experienced persons to the system we needthis algorithm-e proposed algorithm can receive and storethe knowledge of an experienced person and also accu-mulate the experiences of many experienced persons

31 Driversrsquo Experience Ontology In our ontology thedriversrsquo experience class includes two subclasses experi-mental routes and nonexperimental routes -e experi-mental route class encompasses the paths which the

experienced drivers have driven -ese routes are collectedvia an app and are converted to OWL classes -is classitself includes two subclasses traffic and without-trafficclasses -e paths which have been passed at peak traffictimes (7ndash10 AM and 16ndash20 PM) are categorized under thetraffic subclass -e nonexperimental route class encom-passes the paths obtained from the route-finding procedure-ese should also be converted to OWL classes In thedriversrsquo experience ontology for the sake of simplicity and inorder to improve usability only the names of routes aredefined as classes individuals of each class are not defined inthis file For each route class a separate OWL file is definedthe name of which is the same as the corresponding routeclass in the driversrsquo experience ontology -e constituentnodes of a route are stored as individuals in the corre-sponding OWL file using their identification (ID) code as

Any data Any place

Any time

Any network

Any service

Any device

Any user

Ubiquitousinformation

system

Others

Figure 1 Elements of ubiquitous computing

Data collection

Data preparation

Data modeling

Evaluation

Implementation

Monitoring

Ontology designcreating road networks

Predictiveanalytics

Determiningweights

Compare to Google maps based on route length and travel time

Adding new routesto ontology

Figure 2 Research methodology

Complexity 5

their name -e coordinates of each node are defined as dataproperties of individuals in the ontology

-e created ontology is shown in Figure 4 To save spaceonly a limited number of routes are shown here -is figuredepicts the ontology before the route-finding process Afterroute finding if a new path has been created and the timeentered by the user is within the pick range it will be storedin the traffic subclass of the nonexperimental route classOtherwise it will be stored in the nontraffic subclass Eachclass in the lowest level of ontology shows a route and itsname is defined based on its location and destination Forroutes which were traversed on nontraffic times the name ofclass would be location destination and for routes related totraffic classes the name of class would be location_desti-nation_t If the route was traversed on traffic times and it wasretrieved from algorithmrsquos results the name of class wouldbe location_destination_n_t -e flowchart of this ontologyis shown in Figure 5

32 Weighing Method In this paper a graph for a streetnetwork is constructed -e route finding is performed

based on this graph Each segment of the path is consideredas an edge of the graph -e weight of each edge is de-termined based on the number of times the drivers havepassed the corresponding segment Since this number mightbe unexpectedly great its normal form is computed asfollows

fn ei( 1113857 f ei( 1113857 minus fmin + 1fmax minus fmin + 1

(1)

where f(ei) is the number of passes of edge ei and fmin andfmax are the minimum and maximum number of the wholepasses in the graph For a segment which is not located inexperimental routes f(ei) 0 -erefore fmin 0 Gener-ally the weights of the edges of the street network arecomputed using the following equation

w ei( 1113857 maxfn(e)

Length (e)1113888 1113889 minus

fn ei( 1113857

Length ei( 1113857 (2)

where w(ei) denotes the weight of ith edge of the graphLength(ei) is the length of the edge and max(fn

(e)Length(e)) is the maximum number of passes of an edge

Evaluation

Calculating route length and travel time

Combining ontology with the cost model and

applying it to road network

Implementation ofroute-finding algorithm

Data collection

Computing frequency oftraversing a segment Data preprocessing

Determiningweighing method

Converting routes toOWL files

Cost model creation Ontology creation

Figure 3 Workflow of the proposed approach

6 Complexity

divided by the length of that edge Since edges with smallerweights are preferred in Dijkstrarsquos algorithm the values aresubtracted from the maximum value to inverse values

-e reason for choosing Dijkstrarsquos algorithm in thisresearch is that in the proposed algorithm in each exe-cution if a new path is created by an algorithm it will bestored in ontology in order to reuse it in future -ereforewe need to use an algorithm which provides an exact resultwith one hundred percent reliability because the results willbe reused in the future Among the shortest path algo-rithms breadth-first search (BFS) and depth-first search(DFS) can only be used in a weighted graph if the weightsare equal However in our graph weights which are thenumber of traverses by taxis are not equal Moreover theyare slower than in the Dijkstra algorithm [58] -e Greedyalgorithm uses heuristics and Alowast algorithms which are acombination of the Greedy algorithm and the Dijkstraalgorithm -erefore they provide an approximate resultnot an exact result

33 Ubiquitous Ontology-Based Route-Finding AlgorithmFigure 6 depicts the way our proposed algorithm worksFirst the user specifies the location the destination andthe time -en the algorithm determines whether asubclass with the specified location and destination existsin the ontology or not If the time chosen falls within therange of peak times the existing paths in the trafficsubclass are searched Otherwise the paths in the non-traffic subclass are examined When a path exists thestored route is displayed to the user and the algorithmterminates without needing to conduct route findingHowever when there is no corresponding path the al-gorithm determines whether the location and destinationexist in the experimental street network or not If bothexist in the network only the experimental network is usedto construct the graph and find the route (again peak timedetermines the subclass to choosemdashtraffic or nontraffic)

and the algorithm terminates If the location andor thedestination does not exist in the experimental streetnetwork the closest points of the network to them arefound -en a rectangle with a diameter (R2 R1 + 2radic2 h)greater than the distance between the location (or desti-nation) and the point is extracted from the main roadnetwork -e flowchart of the proposed hierarchical al-gorithm and the extraction of the rectangular region fromthe road network are shown in Figure 7 To clarify theroute-finding procedure consider a situation where boththe location and the destination do not exist in the ex-perimental network In such a situation the route findingincludes three stages (1) route finding from the location toits closest point in the experimental network (SSrsquo) (2)route finding from this point to the closest point of theexperimental network with respect to the destination(SrsquoDrsquo) and (3) route finding from the closest point of theexperimental network (with respect to the destination) tothe destination (DrsquoD) After route finding the resultingroute is stored in the driversrsquo experience ontology as aseparate OWL file Now requests with these locations anddestinations are responded to in real time without re-quiring additional route finding

4 Implementation

In this paper an ontology-based route-finding algorithmbased on the driversrsquo experience in ubiquitous GIS spacehas been proposed -e steps of implementing the algo-rithm are shown in Figure 8 In order to implement theproposed algorithm the required data have been collectedvia an app that stores taxi vehiclesrsquo routes throughOpenStreetMap (OSM) maps -ese routes and theircorresponding pick-up and drop-off stations are located inTehran Iran and are entered into the app by the drivers-en they are converted to shapefiles Preprocessing tasksare performed in the ArcGIS software package and theresults are stored in the Oracle database Each route is also

Figure 4 Driversrsquo experience ontology (to save space only a limited number of routes are shown here)

Complexity 7

converted to an OWL file using OWL Application Pro-gramming Interface (API) in Java Using these routes adriversrsquo experience ontology is separately constructed inProtege software A sample of a route stored in the app isshown in Figure 9

After data preparation the presented algorithm wasimplemented In each run of the app if a new route has beengenerated it will be stored as a class in the driversrsquo expe-rience ontology to be used in future requests without needfor additional processing As a result the routes are stored ina low-volume manner while data processing reduces witheven more runs of the algorithm Each route class in theontology occupies about 20ndash25 percent of its correspondingshapefile Moreover in the presented algorithm two roadnetworks existmdasha peak-time road network and a low-traffic

road network During peak times finding an optimum routebecomes important therefore the algorithm will utilizeDijkstrarsquos algorithm and present the optimum path to theuser

5 Evaluation

In order to evaluate the proposed method 10 different lo-cation-destination pairs covering most of Tehran city wereselected First the experimental road network was fullyconstructed (ER) -en it was constructed only consideringthe low-traffic experimental routes (ETR) Also the shortestpaths between the location-destinations were obtained usingDijkstrarsquos algorithm (DR) For all obtained routes the routelengths were calculated Using driversrsquo reports the travel

Collecting drivers routeswith app

Adding the name (location_destination) of

each new route as a subclass of nontraffic class of experimental

routes

Consider the segments of each path as individuals of

that classrsquo path

Considering the fields (X coordinates Y coordinates)

as the data property

Driversrsquo experienceontology

Experimental route classNonexperimental route class

Adding the name (location_destination) of

each new route as a subclass of nontraffic

class ofnonexperimental routes

Creating experimentaland nonexperimental

route classes

Storing each path in a separate OWL file as a class

Retrieving each class ifrequested by the user

Traffic class of experimental routes

Nontraffic class of experimental routes

Nontraffic class ofnonexperimental routes

Traffic class ofnonexperimental routes

Adding the name (location_destination_t)of each new route as a subclass of traffic class of experimental routes

Is the enteredtime by user is in range

of pick traffic times

Adding the name (location_destination_t)of each new route as a subclass of traffic class

of nonexperimental routes

No Yes

Figure 5 Flowchart of creating driversrsquo experience ontology

8 Complexity

times were also calculated Figure 10 illustrates the resultingroutes for a location-destination pair (Route 10) using thethree methods above

51 Evaluation Based on the Route Length Figure 11 shows adiagram of the length of the paths for the three methods-eroute lengths have been computed by accumulating their

Get the locationdestination and time

from user

Is anyroute with this locationand destination in the

nontraffic classes of driverrsquosexperience ontology

Showing the saved pathto the user

End

Are locationand destinationin experimentalroadnetwork

Using only experimentalroad network for route

finding

Using Dijkstrarsquosalgorithm for route

finding

Finding the nearest point of theexperimental road network to the location

and extraction of a rectangle from themain road network with diameter largerthan the distance between the location

and the nearest point in the experimentalroad network

Is location in experimental

road network

Is destination inexperimental

road network

Route finding with experimentalroad network and segments inthe rectangle(s) of main road

network

Storing route name as a subclass oftraffic class of nonexperimental

road network in driverrsquosexperiences ontology

Finding the nearest point of theexperimental road network to the

destination and extraction of a rectanglefrom the main road network with

diameter larger than the distance betweenthe destination and the nearest point in

the experimental road network

Storing route in aseparate OWL file with

its segments andproperties

Yes

Yes

No

NoNoYes

Is the entered

time in peaktraffic times

Is anyroute with this locationand destination in the

traffic classes of driverrsquosexperience ontology

Considering experimental roadnetwork according to trafficclass of experimental routes

Considering experimental roadnetwork according to nontraffic

class of experimental routes

No Yes

NoNo

Is entered time inpeak traffic times

Storing route name as a subclass ofnontraffic class of nonexperimental

road network indriverrsquos experiences ontology

Yes No

Yes

No

Figure 6 Proposed algorithm flowchart

Complexity 9

constituting segments Figure 12 depicts the ratio of routelengths obtained from the two proposed methods to that ofDijkstrarsquos algorithm Figure 13 compares the mean length ofthe 10 routes for each method

It is observable from the figures that route lengths in theshortest path algorithm are shorter than that of the other twomethods -ey have in contrast greater values when con-sidering high-traffic routes -e mean length of routes is1102 km for Dijkstrarsquos algorithm 1422 km when only con-sidering traffic routes and 134 km when considering all theexperimental (traffic and nontraffic) routes -is is due to taxidrivers who generally select the longest but simultaneously

fastest routes On the other hand Dijkstrarsquos algorithm doesnot consider travel time

52 Evaluation Based on Travel Time Figure 14 shows adiagram of the travel times for the three methods -etravel times have been obtained from different driverswho have driven these routes in peak times and in realconditions Figure 15 depicts the travel times obtainedfrom the two proposed methods to that of Dijkstrarsquos al-gorithm Figure 16 compares the mean travel times foreach method

Experimental road network layer

Basic road network layer

Drsquo

Drsquo

D

Srsquo

SrsquoS

(a)

S

R1

R2

h

D

(b)

Figure 7 (a) Hierarchical experimental road network (b) Extraction of rectangular region from main road network [39]

Trace preprocessing

Ontology design

Ontology-based route finding

Ubiquitous ontology-based route finding

User interface Storingroute if anew routeis created

OWL route filesUser Ontologicalroute-finding

algorithm

GIS and ontologyexperts

Driversrsquo experiences ontology OWL route files Ontology data

Taxiroutedata

Mobile appfor data

collection

Experiencedtaxi drivers

Examiningfrequency ofusing a road

segment

Costmodel

a7

149

10

2 116

15

b

c d

ef

Routenetwork

graph

Figure 8 Steps for implementing the proposed algorithm

10 Complexity

Figure 9 A sample of a stored route in the app which collects data

DR

ETR

ERLocationdestination

Final map

N

S

W E

Figure 10 Resulting routes for a location-destination pair (Route 10) using the three methods

DijkstraExperimental routesExperimental traffic routes

2 3 4 5 6 7 8 9 101Route number

02468

101214161820

Rout

e len

gth

(km

)

Figure 11 Comparison of route length for 10 evaluated routes of the three methods

Complexity 11

It is observable from the figures that travel timesobtained from the hierarchical experimental methodwhen only considering high-traffic conditions are lowesthowever the travel time of the shortest path is the highest-e mean travel time is 80 minutes for Dijkstrarsquos algo-rithm 61 minutes when only considering the high-trafficroutes and 775 minutes when considering all of theexperimental routes Although the routes are long whenconsidering high-traffic routes they are fast due to theselection of low-traffic routes in this case For examplewhen going to Tajrish Square from Valiasr SquareDijkstrarsquos algorithm suggests driving through ValiasrStreet which corresponds to the shortest but not fastestroute However the proposed ontology-based approachsuggests that the driver proceeds mainly along ModarresHighway which reduces the travel time by 35 minutes inhigh-traffic conditions

Another approach to evaluate the two proposedmethodsagainst Dijkstrarsquos algorithm is to find the ideal route amongthem In order to find the ideal route one can divide theroute length by its travel time Obviously when the routelengths for all routes or just two of them are equal the onewhich has the shortest travel time should be selected as theoptimum path On the other hand when the travel times areequal the one which has the shortest length should be se-lected as the optimum path Accordingly in eight cases (of10 cases) the ontology-based route finding considering onlyhigh-traffic routes (ETR) predominates In the other twocases the travel time was roughly equal however the routelengths obtained fromDijkstrarsquos algorithm had lower valuesTo determine the deviation of each method from the idealcase which is shown in Figure 17 the following formula isused

Fj travel length of method j

travel time of method j

j ETR ER Dikjestra

Fe ideal route fraction

Dj Fj minus Fe1113872 1113873

Fe

(3)

-erefore according to the evaluations above it can beclaimed that eliminating the experimental high-traffic routesin high-traffic conditions and utilizing the driversrsquo

02468

10121416

Mea

n of

rout

e len

gth

(km

)

Dijkstra Experimentalroutes

Method name

Experimentaltrafficroutes

Figure 13 Mean of route length for 3 methods

0

20

40

60

80

100

120

Trav

el ti

me (

min

)

1 2 3 4 5 6 7 8 9 10Route number

DijkstraExperimental routesExperimental traffic routes

Figure 14 Comparison of travel time for 10 evaluated routes of 3methods

002040608

112

1 2 3 4 5 6 7 8 9 10

ERDijkstraETRDijkstra

Figure 15 Comparison of travel time for the 2 modes of theproposed method with Dijkstrarsquos algorithm

Dijkstra020406080

Mea

n of

trav

el ti

me (

min

)

Experimentalroutes

Method name

Experimentaltrafficroutes

Figure 16 Mean of travel time for 3 methods

0

05

1

15

2

1 2 3 4 5 6 7 8 9 10

ERDijkstraETRDijkstra

Figure 12 Comparison of route length in 2 modes of the proposedalgorithm with Dijkstrarsquos algorithm

12 Complexity

experiences make the algorithm generate the optimum routethrough low-traffic parts of the network -is minimizestravel time which is the most important parameter in routefinding

6 Conclusions

Different people gain different spatial experiences Asthere are no systems to store these experiences theyevaporate over time-e present study provides the abilityto store spatial experiences and to make the existingsystems in ubiquitous GIS space intelligent In this paperan ontology-based route-finding algorithm in ubiquitousGIS space was designed and implemented using theroutes driven by the drivers of Tehran in high- and low-traffic conditions

In the proposed route-finding algorithm based on thedriversrsquo experiences ontology the number of times that eachsegment is passed by drivers is used to assign weights to thatsegment of the network First the user specifies the location thedestination and the time -en the algorithm determineswhether a subclass with the specified location and destinationexists in the ontology or not If the time chosen falls duringpeak times the existing paths in the traffic subclass are searchedthrough Otherwise the paths in the nontraffic subclass areexamined When a path exists the stored route is displayed tothe user and the algorithm terminates without needing toconduct route finding However when there is no corre-sponding path the algorithm determines whether the locationand destination exist in the experimental road network or notIf both exist in the network only the experimental is used toconstruct the graph and find the route (again peak time

determines the subclass to choosemdashtraffic or nontraffic) andthe algorithm terminates If the origin andor the destinationdoes not exist in the experimental road network the closestpoints of the network to them are found-en a rectangle witha diameter (R2R1 + 2radic2h) greater than the distance betweenthe location (or destination) and the point is extracted from themain road network -e main difference of this algorithm tothe previous experimental route-finding algorithms is the useof ontology and the construction of road networks onlyconsidering the low-traffic paths when there are high-trafficpaths in the network Since each newpath is stored as a separateOWL class the storage issues in the databases are resolved Inaddition the use of ontology causes the system to update everytime the algorithm runs and generates new paths Utilizing onlylow-traffic routes in high-traffic conditions allows the system tosuggest routes that are faster ensuring the optimum path basedon travel time is achieved According to our evaluations theproposed algorithm suggests routes that are longer but si-multaneously the fastest -is guarantees the appropriatenessof using such an algorithm in route finding In this researchany user can access the ontology from any location and at anytime to conduct route finding Moreover being intelligent theubiquitous environment can receive and store the experiences(any data element of ubiquitous GIS) of any person to utilizewithin its services in the future

More innovative applications of spatial experiencemodeling such as ambulance firefighting and tourismservices can be subjects for further considerations -eprocess of route finding for this category is also affected bytraffic as emergency services can also become stuck in trafficlike other drivers because there are no dedicated lines on allroads Due to the limitation in length of road and a largenumber of vehicles in the road in traffic times it is hard forother vehicles to provide a route for traversing of the am-bulance and they still need to find a way to avoid traffic-erefore the experiences from expert technicians in am-bulances or firefighters who have been in real event con-ditions can be modeled and used in the future In additionthis study considers only two factors distance and time forevaluation Other criteria including fuel consumption can beconsidered in future

Data Availability

-e authors are ready to share the research data on request

Disclosure

Maryam Barzegar and Abolghasem Sadeghi-Niaraki are tobe considered as co-first authors

Conflicts of Interest

-e authors declare that they have no conflicts of interest

Acknowledgments

-is research was supported by theMSIT (Ministry of Scienceand ICT) Korea under the ITRC (Information Technology

0

01

02

03

041

2

3

4

5

6

7

8

9

10

DijkstraExperimental routes

Experimental traffic routesIdeal

Figure 17 Comparison of routes from the 3methods with the idealroute

Complexity 13

Research Center) support program (IITP-2019-2016-0-00312)supervised by the IITP (Institute for Information andCommunications Technology Planning and Evaluation)

References

[1] I P Tussyadiah and F J Zach ldquo-e role of geo-basedtechnology in place experiencesrdquo Annals of Tourism Researchvol 39 no 2 pp 780ndash800 2012

[2] G M Honti and J Abonyi ldquoA review of semantic sensortechnologies in internet of things architecturesrdquo Complexityvol 2019 Article ID 6473160 21 pages 2019

[3] U H Govindarajan A J Trappey and C V TrappeyldquoImmersive technology for human-centric cyberphysicalsystems in complex manufacturing processes a compre-hensive overview of the global patent profile using collectiveintelligencerdquo Complexity vol 2018 Article ID 428363417 pages 2018

[4] B K Foguem T Coudert C Beler and L GenesteldquoKnowledge formalization in experience feedback processesan ontology-based approachrdquo Computers in Industry vol 59no 7 pp 694ndash710 2008

[5] N Lasierra F Roldan A Alesanco and J Garcıa ldquoTowardsimproving usage and management of supplies in healthcarean ontology-based solution for sharing knowledgerdquo ExpertSystems with Applications vol 41 no 14 pp 6261ndash6273 2014

[6] M-H Abel ldquoKnowledge map-based web platform to facilitateorganizational learning return of experiencesrdquo Computers inHuman Behavior vol 51 pp 960ndash966 2015

[7] A Garcıa-Crespo J L Lopez-Cuadrado R Colomo-PalaciosI Gonzalez-Carrasco and B Ruiz-Mezcua ldquoSem-fit a se-mantic based expert system to provide recommendations inthe tourism domainrdquo Expert Systems with Applicationsvol 38 no 10 pp 13310ndash13319 2011

[8] D Mourtzis M Doukas and C Giannoulis ldquoAn inference-based knowledge reuse framework for historical product andproduction information retrievalrdquo Procedia CIRP vol 41pp 472ndash477 2016

[9] K Efthymiou K Sipsas D Mourtzis and G ChryssolourisldquoOn knowledge reuse for manufacturing systems design andplanning a semantic technology approachrdquo CIRP Journal ofManufacturing Science and Technology vol 8 pp 1ndash11 2015

[10] W L Mikos J C E Ferreira P E A Botura and L S FreitasldquoA system for distributed sharing and reuse of design andmanufacturing knowledge in the PFMEA domain using adescription logics-based ontologyrdquo Journal of ManufacturingSystems vol 30 no 3 pp 133ndash143 2011

[11] P P Ruiz B K Foguem and B Grabot ldquoGeneratingknowledge in maintenance from experience feedbackrdquoKnowledge-Based Systems vol 68 pp 4ndash20 2014

[12] S Moehrle and W Raskob ldquoStructuring and reusingknowledge from historical events for supporting nuclearemergency and remediation managementrdquo Engineering Ap-plications of Artificial Intelligence vol 46 pp 303ndash311 2015

[13] Q Meng Z Zhang X Wan and X Rong ldquoProperties ex-ploring and information mining in consumer communitynetwork a case of huawei pollen clubrdquo Complexity vol 2018Article ID 9470580 19 pages 2018

[14] R S Renu and GMocko ldquoComputing similarity of text-basedassembly processes for knowledge retrieval and reuserdquoJournal of Manufacturing Systems vol 39 pp 101ndash110 2016

[15] G E Modoni M Doukas W Terkaj M Sacco andD Mourtzis ldquoEnhancing factory data integration through thedevelopment of an ontology from the reference models reuse

to the semantic conversion of the legacy modelsrdquo In-ternational Journal of Computer Integrated Manufacturingvol 30 no 10 pp 1043ndash1059 2017

[16] E R Reyes S Negny G C Robles and J M Le LannldquoImprovement of online adaptation knowledge acquisitionand reuse in case-based reasoning application to processengineering designrdquo Engineering Applications of ArtificialIntelligence vol 41 pp 1ndash16 2015

[17] B Kamsu-Foguem and F H Abanda ldquoExperience modelingwith graphs encoded knowledge for construction industryrdquoComputers in Industry vol 70 pp 79ndash88 2015

[18] B Kamsu-Foguem F H Abanda M B Doumbouya andJ F Tchouanguem ldquoGraph-based ontology reasoning forformal verification of BREEAM rulesrdquo Cognitive SystemsResearch vol 55 pp 14ndash33 2019

[19] S H Othman and G Beydoun ldquoA metamodel-basedknowledge sharing system for disaster managementrdquo ExpertSystems with Applications vol 63 pp 49ndash65 2016

[20] D Xia X Lu H Li W Wang Y Li and Z Zhang ldquoAmapreduce-based parallel frequent pattern growth algorithmfor spatiotemporal association analysis of mobile trajectorybig datardquo Complexity vol 2018 Article ID 2818251 16 pages2018

[21] J Geng S Wang W Gan et al ldquoPromoting geospatial servicefrom information to knowledge with spatiotemporal se-manticsrdquo Complexity vol 2019 Article ID 9301420 14 pages2019

[22] A Nowak-Brzezinska ldquoEnhancing the efficiency of a decisionsupport system through the clustering of complex rule-basedknowledge bases and modification of the inference algo-rithmrdquo Complexity vol 2018 Article ID 2065491 14 pages2018

[23] E Maleki F Belkadi N Boli et al ldquoOntology-basedframework enabling smart product-service systems appli-cation of sensing systems for machine health monitoringrdquoIEEE Internet of ings Journal vol 5 no 6 pp 4496ndash45052018

[24] F H Abanda B Kamsu-Foguem and J H M TahldquoBIMmdashnew rules of measurement ontology for constructioncost estimationrdquo Engineering Science and Technology anInternational Journal vol 20 no 2 pp 443ndash459 2017

[25] B Kamsu-Foguem and P Tiako ldquoRisk information formal-isation with graphsrdquo Computers in Industry vol 85 pp 58ndash69 2017

[26] M Barzegar A Sadeghi-Niaraki and M Shakeri ldquoSpatialexperience based route finding using ontologiesrdquo ETRIJournal pp 1ndash11 2019

[27] E Camossi P Villa and L Mazzola ldquoSemantic-basedanomalous pattern discovery in moving object trajectoriesrdquo2013 httpsarxivorgabs13051946

[28] R Wannous J Malki A Bouju and C Vincent ldquoModellingmobile object activities based on trajectory ontology rulesconsidering spatial relationship rulesrdquo in Modeling Ap-proaches and Algorithms for Advanced Computer Applicationspp 249ndash258 Springer International Publishing BerlinGermany 2013

[29] Y Hu K Janowicz D Carral et al ldquoA geo-ontology designpattern for semantic trajectoriesrdquo in Proceedings of the In-ternational Conference on Spatial Information eorypp 438ndash456 Springer International Publishing ScarboroughUK September 2013

[30] M Baglioni J Macedo C Renso and M Wachowicz ldquoAnontology-based approach for the semantic modelling andreasoning on trajectoriesrdquo in Advances in Conceptual

14 Complexity

can be used in route-finding algorithms-e general trend ofthis paper is that first ontology is designed for route finding-en the data needed for this ontology (paths for taxidrivers) are collected using an application Next based onthe number of times that each segment is passed by driversthe cost of each segment of the path is determined and anontology-based route finding algorithm of the driversrsquo ex-periences is proposed Finally the proposed method isimplemented in a ubiquitous GIS for the city of Tehran andis compared to Dijkstrarsquos algorithm in terms of travel timeand route length It should be noted that improving route-finding algorithm using ontology-based modeling of driverrsquospatial experiences facilitates the timeliness of servicesprovides the capability to make use of earlier informationwithout reprocessing and enhances sharing reusing andprocessing domain knowledge [48 49] -e structure of thispaper is as follows In Section 2 the concepts of ubiquitousGIS and the cause of using ontology for modeling in-dividualrsquos experiences in ubiquitous GIS are explained InSection 3 the general method of this paper driversrsquo expe-riences ontology weighting method and the proposedontology-based route-finding algorithm are introduced -eproposed algorithm is implemented using Tehranrsquos data inSection 4 In Section 5 the proposed method is compared toDijkstrarsquos algorithm in terms of travel time and route lengthFinally general conclusions of the proposed method and itsadvantages are presented

2 Ubiquitous Computing

With the growth of science and technology new generationsof computing have appeared Ubiquitous computing whichis a new generation of computing systems includes a vastrange of computers and small sensors embedded in theenvironment-e goal of ubiquitous computing is in fact todiminish the virtual dimension of technology in our lives Inubiquitous computing all the limitations are removed forhuman beings Figure 1 illustrates the elements of ubiquitouscomputing In ubiquitous computing users can access theirdesired services in any location at any time using anydevice and without any limitations in the network At thesame time the huge volume of information and services intodayrsquos world have caused the introduction of some newconcepts such as providing intelligent location-based ser-vices under any situation at any time and location using anydevice and service via any network and for any userConsidering the close relationships between GIS and in-formation technology (IT) geomatics especially GIS fol-lowed this trend and introduced ubiquitous GeospatialInformation Systems Ubiquitous GIS is based on mobilecomputing technology environment and provides mobileand distributed geographic information services It in-tegrates GIS GPS and wireless communication [50]Ubiquitous GIS enhances the provision of spatial data andservices to users even public who have no professionalknowledge of GIS in an easy-to-use way [51 52]

-e major issues with the exact route-finding algorithmsare the relatively large time it takes to find the route betweentwo locations and the high volume of data stored in a

database In a ubiquitous environment the system needs torespond to the userrsquos requests in a real-time manner hencethe noted issues are barriers that should be resolved In thispaper the ontologies are utilized to overcome this issue toenhance knowledge sharing and reusing between differentusers In other words ontologies help the stored knowledgeto be used by different users many times In this case anyuser element of the ubiquitous environment is supportedOntologies provide the capability to make use of earlierinformation without reprocessing [53] -is facilitates thetimeliness of services In addition using ontology in route-finding algorithm was to store the knowledge of experiencedpersons -is makes the systems smart ubiquitous and actlike an experienced human Moreover the data in ontologiesare stored in plain text formats such as Web OntologyLanguage (OWL) Obviously plain text formats requiremuch lower storage space Moreover ontologies are in-dependent from text and implementation and operate on ahigher level of abstraction However databases are situatedat the lower level of abstraction and they are mainly designedto meet the requirements of a particular application orcorporation and when the requirements change the schemaof a database also need to be modified [54 55] In our re-search since we update the ontology many times relationaldatabases cannot be used Relational databases are not ap-propriate for applications that need to make frequent up-dates since you need to change the schema for each updateIn addition ontologies do not need to be normalized (suchcharacteristics make for easier the information sharing andmerging) [56] However relational databases need to benormalized Besides ontology languages are more expressivein terms of expressing more semantic concepts than data-base languages which only include constructs for defining orextracting data [55] In addition ontology languages providea more correct and precise domain conceptualization [57]

In this research any user can access the ontology at anytime and in any location to find routes In addition due tothe intelligence of ubiquitous environment the experiencesfrom users (any data element of ubiquitous computing) canbe received and stored to be utilized for future services

3 Methodology

Figure 2 shows the main processes of this research Based onthese processes the ontology-based route-finding algorithmis proposed which enables to store and retrieve spatial ex-periences of the taxi drivers

An ontology-based algorithm for storing spatial expe-riences in ubiquitous GIS space has been presented Figure 3depicts the workflow of the proposed approach First thedataset is collected via an app -is app collects the routes oftaxi vehicles needed for the ontology -e number of timesthat each segment is passed by drivers is calculated and usedto weight the edges of the street network and to build the costmodel -en taxi paths are converted to the driversrsquo ex-perience class in the ontology Finally an ontology-basedroute-finding algorithm is designed implemented andevaluated in ubiquitous environment -e work has beendescribed in detail in the following sections

4 Complexity

-e algorithm makes the route-finding systems smartand ubiquitous intended to act like experienced people Aubiquitous system is a smart system which acts not only likea person but also like an experienced human A person whoworks for some field of interest for many years gets a lot ofexperience-is person has much knowledge To transfer theknowledge of the experienced persons to the system we needthis algorithm-e proposed algorithm can receive and storethe knowledge of an experienced person and also accu-mulate the experiences of many experienced persons

31 Driversrsquo Experience Ontology In our ontology thedriversrsquo experience class includes two subclasses experi-mental routes and nonexperimental routes -e experi-mental route class encompasses the paths which the

experienced drivers have driven -ese routes are collectedvia an app and are converted to OWL classes -is classitself includes two subclasses traffic and without-trafficclasses -e paths which have been passed at peak traffictimes (7ndash10 AM and 16ndash20 PM) are categorized under thetraffic subclass -e nonexperimental route class encom-passes the paths obtained from the route-finding procedure-ese should also be converted to OWL classes In thedriversrsquo experience ontology for the sake of simplicity and inorder to improve usability only the names of routes aredefined as classes individuals of each class are not defined inthis file For each route class a separate OWL file is definedthe name of which is the same as the corresponding routeclass in the driversrsquo experience ontology -e constituentnodes of a route are stored as individuals in the corre-sponding OWL file using their identification (ID) code as

Any data Any place

Any time

Any network

Any service

Any device

Any user

Ubiquitousinformation

system

Others

Figure 1 Elements of ubiquitous computing

Data collection

Data preparation

Data modeling

Evaluation

Implementation

Monitoring

Ontology designcreating road networks

Predictiveanalytics

Determiningweights

Compare to Google maps based on route length and travel time

Adding new routesto ontology

Figure 2 Research methodology

Complexity 5

their name -e coordinates of each node are defined as dataproperties of individuals in the ontology

-e created ontology is shown in Figure 4 To save spaceonly a limited number of routes are shown here -is figuredepicts the ontology before the route-finding process Afterroute finding if a new path has been created and the timeentered by the user is within the pick range it will be storedin the traffic subclass of the nonexperimental route classOtherwise it will be stored in the nontraffic subclass Eachclass in the lowest level of ontology shows a route and itsname is defined based on its location and destination Forroutes which were traversed on nontraffic times the name ofclass would be location destination and for routes related totraffic classes the name of class would be location_desti-nation_t If the route was traversed on traffic times and it wasretrieved from algorithmrsquos results the name of class wouldbe location_destination_n_t -e flowchart of this ontologyis shown in Figure 5

32 Weighing Method In this paper a graph for a streetnetwork is constructed -e route finding is performed

based on this graph Each segment of the path is consideredas an edge of the graph -e weight of each edge is de-termined based on the number of times the drivers havepassed the corresponding segment Since this number mightbe unexpectedly great its normal form is computed asfollows

fn ei( 1113857 f ei( 1113857 minus fmin + 1fmax minus fmin + 1

(1)

where f(ei) is the number of passes of edge ei and fmin andfmax are the minimum and maximum number of the wholepasses in the graph For a segment which is not located inexperimental routes f(ei) 0 -erefore fmin 0 Gener-ally the weights of the edges of the street network arecomputed using the following equation

w ei( 1113857 maxfn(e)

Length (e)1113888 1113889 minus

fn ei( 1113857

Length ei( 1113857 (2)

where w(ei) denotes the weight of ith edge of the graphLength(ei) is the length of the edge and max(fn

(e)Length(e)) is the maximum number of passes of an edge

Evaluation

Calculating route length and travel time

Combining ontology with the cost model and

applying it to road network

Implementation ofroute-finding algorithm

Data collection

Computing frequency oftraversing a segment Data preprocessing

Determiningweighing method

Converting routes toOWL files

Cost model creation Ontology creation

Figure 3 Workflow of the proposed approach

6 Complexity

divided by the length of that edge Since edges with smallerweights are preferred in Dijkstrarsquos algorithm the values aresubtracted from the maximum value to inverse values

-e reason for choosing Dijkstrarsquos algorithm in thisresearch is that in the proposed algorithm in each exe-cution if a new path is created by an algorithm it will bestored in ontology in order to reuse it in future -ereforewe need to use an algorithm which provides an exact resultwith one hundred percent reliability because the results willbe reused in the future Among the shortest path algo-rithms breadth-first search (BFS) and depth-first search(DFS) can only be used in a weighted graph if the weightsare equal However in our graph weights which are thenumber of traverses by taxis are not equal Moreover theyare slower than in the Dijkstra algorithm [58] -e Greedyalgorithm uses heuristics and Alowast algorithms which are acombination of the Greedy algorithm and the Dijkstraalgorithm -erefore they provide an approximate resultnot an exact result

33 Ubiquitous Ontology-Based Route-Finding AlgorithmFigure 6 depicts the way our proposed algorithm worksFirst the user specifies the location the destination andthe time -en the algorithm determines whether asubclass with the specified location and destination existsin the ontology or not If the time chosen falls within therange of peak times the existing paths in the trafficsubclass are searched Otherwise the paths in the non-traffic subclass are examined When a path exists thestored route is displayed to the user and the algorithmterminates without needing to conduct route findingHowever when there is no corresponding path the al-gorithm determines whether the location and destinationexist in the experimental street network or not If bothexist in the network only the experimental network is usedto construct the graph and find the route (again peak timedetermines the subclass to choosemdashtraffic or nontraffic)

and the algorithm terminates If the location andor thedestination does not exist in the experimental streetnetwork the closest points of the network to them arefound -en a rectangle with a diameter (R2 R1 + 2radic2 h)greater than the distance between the location (or desti-nation) and the point is extracted from the main roadnetwork -e flowchart of the proposed hierarchical al-gorithm and the extraction of the rectangular region fromthe road network are shown in Figure 7 To clarify theroute-finding procedure consider a situation where boththe location and the destination do not exist in the ex-perimental network In such a situation the route findingincludes three stages (1) route finding from the location toits closest point in the experimental network (SSrsquo) (2)route finding from this point to the closest point of theexperimental network with respect to the destination(SrsquoDrsquo) and (3) route finding from the closest point of theexperimental network (with respect to the destination) tothe destination (DrsquoD) After route finding the resultingroute is stored in the driversrsquo experience ontology as aseparate OWL file Now requests with these locations anddestinations are responded to in real time without re-quiring additional route finding

4 Implementation

In this paper an ontology-based route-finding algorithmbased on the driversrsquo experience in ubiquitous GIS spacehas been proposed -e steps of implementing the algo-rithm are shown in Figure 8 In order to implement theproposed algorithm the required data have been collectedvia an app that stores taxi vehiclesrsquo routes throughOpenStreetMap (OSM) maps -ese routes and theircorresponding pick-up and drop-off stations are located inTehran Iran and are entered into the app by the drivers-en they are converted to shapefiles Preprocessing tasksare performed in the ArcGIS software package and theresults are stored in the Oracle database Each route is also

Figure 4 Driversrsquo experience ontology (to save space only a limited number of routes are shown here)

Complexity 7

converted to an OWL file using OWL Application Pro-gramming Interface (API) in Java Using these routes adriversrsquo experience ontology is separately constructed inProtege software A sample of a route stored in the app isshown in Figure 9

After data preparation the presented algorithm wasimplemented In each run of the app if a new route has beengenerated it will be stored as a class in the driversrsquo expe-rience ontology to be used in future requests without needfor additional processing As a result the routes are stored ina low-volume manner while data processing reduces witheven more runs of the algorithm Each route class in theontology occupies about 20ndash25 percent of its correspondingshapefile Moreover in the presented algorithm two roadnetworks existmdasha peak-time road network and a low-traffic

road network During peak times finding an optimum routebecomes important therefore the algorithm will utilizeDijkstrarsquos algorithm and present the optimum path to theuser

5 Evaluation

In order to evaluate the proposed method 10 different lo-cation-destination pairs covering most of Tehran city wereselected First the experimental road network was fullyconstructed (ER) -en it was constructed only consideringthe low-traffic experimental routes (ETR) Also the shortestpaths between the location-destinations were obtained usingDijkstrarsquos algorithm (DR) For all obtained routes the routelengths were calculated Using driversrsquo reports the travel

Collecting drivers routeswith app

Adding the name (location_destination) of

each new route as a subclass of nontraffic class of experimental

routes

Consider the segments of each path as individuals of

that classrsquo path

Considering the fields (X coordinates Y coordinates)

as the data property

Driversrsquo experienceontology

Experimental route classNonexperimental route class

Adding the name (location_destination) of

each new route as a subclass of nontraffic

class ofnonexperimental routes

Creating experimentaland nonexperimental

route classes

Storing each path in a separate OWL file as a class

Retrieving each class ifrequested by the user

Traffic class of experimental routes

Nontraffic class of experimental routes

Nontraffic class ofnonexperimental routes

Traffic class ofnonexperimental routes

Adding the name (location_destination_t)of each new route as a subclass of traffic class of experimental routes

Is the enteredtime by user is in range

of pick traffic times

Adding the name (location_destination_t)of each new route as a subclass of traffic class

of nonexperimental routes

No Yes

Figure 5 Flowchart of creating driversrsquo experience ontology

8 Complexity

times were also calculated Figure 10 illustrates the resultingroutes for a location-destination pair (Route 10) using thethree methods above

51 Evaluation Based on the Route Length Figure 11 shows adiagram of the length of the paths for the three methods-eroute lengths have been computed by accumulating their

Get the locationdestination and time

from user

Is anyroute with this locationand destination in the

nontraffic classes of driverrsquosexperience ontology

Showing the saved pathto the user

End

Are locationand destinationin experimentalroadnetwork

Using only experimentalroad network for route

finding

Using Dijkstrarsquosalgorithm for route

finding

Finding the nearest point of theexperimental road network to the location

and extraction of a rectangle from themain road network with diameter largerthan the distance between the location

and the nearest point in the experimentalroad network

Is location in experimental

road network

Is destination inexperimental

road network

Route finding with experimentalroad network and segments inthe rectangle(s) of main road

network

Storing route name as a subclass oftraffic class of nonexperimental

road network in driverrsquosexperiences ontology

Finding the nearest point of theexperimental road network to the

destination and extraction of a rectanglefrom the main road network with

diameter larger than the distance betweenthe destination and the nearest point in

the experimental road network

Storing route in aseparate OWL file with

its segments andproperties

Yes

Yes

No

NoNoYes

Is the entered

time in peaktraffic times

Is anyroute with this locationand destination in the

traffic classes of driverrsquosexperience ontology

Considering experimental roadnetwork according to trafficclass of experimental routes

Considering experimental roadnetwork according to nontraffic

class of experimental routes

No Yes

NoNo

Is entered time inpeak traffic times

Storing route name as a subclass ofnontraffic class of nonexperimental

road network indriverrsquos experiences ontology

Yes No

Yes

No

Figure 6 Proposed algorithm flowchart

Complexity 9

constituting segments Figure 12 depicts the ratio of routelengths obtained from the two proposed methods to that ofDijkstrarsquos algorithm Figure 13 compares the mean length ofthe 10 routes for each method

It is observable from the figures that route lengths in theshortest path algorithm are shorter than that of the other twomethods -ey have in contrast greater values when con-sidering high-traffic routes -e mean length of routes is1102 km for Dijkstrarsquos algorithm 1422 km when only con-sidering traffic routes and 134 km when considering all theexperimental (traffic and nontraffic) routes -is is due to taxidrivers who generally select the longest but simultaneously

fastest routes On the other hand Dijkstrarsquos algorithm doesnot consider travel time

52 Evaluation Based on Travel Time Figure 14 shows adiagram of the travel times for the three methods -etravel times have been obtained from different driverswho have driven these routes in peak times and in realconditions Figure 15 depicts the travel times obtainedfrom the two proposed methods to that of Dijkstrarsquos al-gorithm Figure 16 compares the mean travel times foreach method

Experimental road network layer

Basic road network layer

Drsquo

Drsquo

D

Srsquo

SrsquoS

(a)

S

R1

R2

h

D

(b)

Figure 7 (a) Hierarchical experimental road network (b) Extraction of rectangular region from main road network [39]

Trace preprocessing

Ontology design

Ontology-based route finding

Ubiquitous ontology-based route finding

User interface Storingroute if anew routeis created

OWL route filesUser Ontologicalroute-finding

algorithm

GIS and ontologyexperts

Driversrsquo experiences ontology OWL route files Ontology data

Taxiroutedata

Mobile appfor data

collection

Experiencedtaxi drivers

Examiningfrequency ofusing a road

segment

Costmodel

a7

149

10

2 116

15

b

c d

ef

Routenetwork

graph

Figure 8 Steps for implementing the proposed algorithm

10 Complexity

Figure 9 A sample of a stored route in the app which collects data

DR

ETR

ERLocationdestination

Final map

N

S

W E

Figure 10 Resulting routes for a location-destination pair (Route 10) using the three methods

DijkstraExperimental routesExperimental traffic routes

2 3 4 5 6 7 8 9 101Route number

02468

101214161820

Rout

e len

gth

(km

)

Figure 11 Comparison of route length for 10 evaluated routes of the three methods

Complexity 11

It is observable from the figures that travel timesobtained from the hierarchical experimental methodwhen only considering high-traffic conditions are lowesthowever the travel time of the shortest path is the highest-e mean travel time is 80 minutes for Dijkstrarsquos algo-rithm 61 minutes when only considering the high-trafficroutes and 775 minutes when considering all of theexperimental routes Although the routes are long whenconsidering high-traffic routes they are fast due to theselection of low-traffic routes in this case For examplewhen going to Tajrish Square from Valiasr SquareDijkstrarsquos algorithm suggests driving through ValiasrStreet which corresponds to the shortest but not fastestroute However the proposed ontology-based approachsuggests that the driver proceeds mainly along ModarresHighway which reduces the travel time by 35 minutes inhigh-traffic conditions

Another approach to evaluate the two proposedmethodsagainst Dijkstrarsquos algorithm is to find the ideal route amongthem In order to find the ideal route one can divide theroute length by its travel time Obviously when the routelengths for all routes or just two of them are equal the onewhich has the shortest travel time should be selected as theoptimum path On the other hand when the travel times areequal the one which has the shortest length should be se-lected as the optimum path Accordingly in eight cases (of10 cases) the ontology-based route finding considering onlyhigh-traffic routes (ETR) predominates In the other twocases the travel time was roughly equal however the routelengths obtained fromDijkstrarsquos algorithm had lower valuesTo determine the deviation of each method from the idealcase which is shown in Figure 17 the following formula isused

Fj travel length of method j

travel time of method j

j ETR ER Dikjestra

Fe ideal route fraction

Dj Fj minus Fe1113872 1113873

Fe

(3)

-erefore according to the evaluations above it can beclaimed that eliminating the experimental high-traffic routesin high-traffic conditions and utilizing the driversrsquo

02468

10121416

Mea

n of

rout

e len

gth

(km

)

Dijkstra Experimentalroutes

Method name

Experimentaltrafficroutes

Figure 13 Mean of route length for 3 methods

0

20

40

60

80

100

120

Trav

el ti

me (

min

)

1 2 3 4 5 6 7 8 9 10Route number

DijkstraExperimental routesExperimental traffic routes

Figure 14 Comparison of travel time for 10 evaluated routes of 3methods

002040608

112

1 2 3 4 5 6 7 8 9 10

ERDijkstraETRDijkstra

Figure 15 Comparison of travel time for the 2 modes of theproposed method with Dijkstrarsquos algorithm

Dijkstra020406080

Mea

n of

trav

el ti

me (

min

)

Experimentalroutes

Method name

Experimentaltrafficroutes

Figure 16 Mean of travel time for 3 methods

0

05

1

15

2

1 2 3 4 5 6 7 8 9 10

ERDijkstraETRDijkstra

Figure 12 Comparison of route length in 2 modes of the proposedalgorithm with Dijkstrarsquos algorithm

12 Complexity

experiences make the algorithm generate the optimum routethrough low-traffic parts of the network -is minimizestravel time which is the most important parameter in routefinding

6 Conclusions

Different people gain different spatial experiences Asthere are no systems to store these experiences theyevaporate over time-e present study provides the abilityto store spatial experiences and to make the existingsystems in ubiquitous GIS space intelligent In this paperan ontology-based route-finding algorithm in ubiquitousGIS space was designed and implemented using theroutes driven by the drivers of Tehran in high- and low-traffic conditions

In the proposed route-finding algorithm based on thedriversrsquo experiences ontology the number of times that eachsegment is passed by drivers is used to assign weights to thatsegment of the network First the user specifies the location thedestination and the time -en the algorithm determineswhether a subclass with the specified location and destinationexists in the ontology or not If the time chosen falls duringpeak times the existing paths in the traffic subclass are searchedthrough Otherwise the paths in the nontraffic subclass areexamined When a path exists the stored route is displayed tothe user and the algorithm terminates without needing toconduct route finding However when there is no corre-sponding path the algorithm determines whether the locationand destination exist in the experimental road network or notIf both exist in the network only the experimental is used toconstruct the graph and find the route (again peak time

determines the subclass to choosemdashtraffic or nontraffic) andthe algorithm terminates If the origin andor the destinationdoes not exist in the experimental road network the closestpoints of the network to them are found-en a rectangle witha diameter (R2R1 + 2radic2h) greater than the distance betweenthe location (or destination) and the point is extracted from themain road network -e main difference of this algorithm tothe previous experimental route-finding algorithms is the useof ontology and the construction of road networks onlyconsidering the low-traffic paths when there are high-trafficpaths in the network Since each newpath is stored as a separateOWL class the storage issues in the databases are resolved Inaddition the use of ontology causes the system to update everytime the algorithm runs and generates new paths Utilizing onlylow-traffic routes in high-traffic conditions allows the system tosuggest routes that are faster ensuring the optimum path basedon travel time is achieved According to our evaluations theproposed algorithm suggests routes that are longer but si-multaneously the fastest -is guarantees the appropriatenessof using such an algorithm in route finding In this researchany user can access the ontology from any location and at anytime to conduct route finding Moreover being intelligent theubiquitous environment can receive and store the experiences(any data element of ubiquitous GIS) of any person to utilizewithin its services in the future

More innovative applications of spatial experiencemodeling such as ambulance firefighting and tourismservices can be subjects for further considerations -eprocess of route finding for this category is also affected bytraffic as emergency services can also become stuck in trafficlike other drivers because there are no dedicated lines on allroads Due to the limitation in length of road and a largenumber of vehicles in the road in traffic times it is hard forother vehicles to provide a route for traversing of the am-bulance and they still need to find a way to avoid traffic-erefore the experiences from expert technicians in am-bulances or firefighters who have been in real event con-ditions can be modeled and used in the future In additionthis study considers only two factors distance and time forevaluation Other criteria including fuel consumption can beconsidered in future

Data Availability

-e authors are ready to share the research data on request

Disclosure

Maryam Barzegar and Abolghasem Sadeghi-Niaraki are tobe considered as co-first authors

Conflicts of Interest

-e authors declare that they have no conflicts of interest

Acknowledgments

-is research was supported by theMSIT (Ministry of Scienceand ICT) Korea under the ITRC (Information Technology

0

01

02

03

041

2

3

4

5

6

7

8

9

10

DijkstraExperimental routes

Experimental traffic routesIdeal

Figure 17 Comparison of routes from the 3methods with the idealroute

Complexity 13

Research Center) support program (IITP-2019-2016-0-00312)supervised by the IITP (Institute for Information andCommunications Technology Planning and Evaluation)

References

[1] I P Tussyadiah and F J Zach ldquo-e role of geo-basedtechnology in place experiencesrdquo Annals of Tourism Researchvol 39 no 2 pp 780ndash800 2012

[2] G M Honti and J Abonyi ldquoA review of semantic sensortechnologies in internet of things architecturesrdquo Complexityvol 2019 Article ID 6473160 21 pages 2019

[3] U H Govindarajan A J Trappey and C V TrappeyldquoImmersive technology for human-centric cyberphysicalsystems in complex manufacturing processes a compre-hensive overview of the global patent profile using collectiveintelligencerdquo Complexity vol 2018 Article ID 428363417 pages 2018

[4] B K Foguem T Coudert C Beler and L GenesteldquoKnowledge formalization in experience feedback processesan ontology-based approachrdquo Computers in Industry vol 59no 7 pp 694ndash710 2008

[5] N Lasierra F Roldan A Alesanco and J Garcıa ldquoTowardsimproving usage and management of supplies in healthcarean ontology-based solution for sharing knowledgerdquo ExpertSystems with Applications vol 41 no 14 pp 6261ndash6273 2014

[6] M-H Abel ldquoKnowledge map-based web platform to facilitateorganizational learning return of experiencesrdquo Computers inHuman Behavior vol 51 pp 960ndash966 2015

[7] A Garcıa-Crespo J L Lopez-Cuadrado R Colomo-PalaciosI Gonzalez-Carrasco and B Ruiz-Mezcua ldquoSem-fit a se-mantic based expert system to provide recommendations inthe tourism domainrdquo Expert Systems with Applicationsvol 38 no 10 pp 13310ndash13319 2011

[8] D Mourtzis M Doukas and C Giannoulis ldquoAn inference-based knowledge reuse framework for historical product andproduction information retrievalrdquo Procedia CIRP vol 41pp 472ndash477 2016

[9] K Efthymiou K Sipsas D Mourtzis and G ChryssolourisldquoOn knowledge reuse for manufacturing systems design andplanning a semantic technology approachrdquo CIRP Journal ofManufacturing Science and Technology vol 8 pp 1ndash11 2015

[10] W L Mikos J C E Ferreira P E A Botura and L S FreitasldquoA system for distributed sharing and reuse of design andmanufacturing knowledge in the PFMEA domain using adescription logics-based ontologyrdquo Journal of ManufacturingSystems vol 30 no 3 pp 133ndash143 2011

[11] P P Ruiz B K Foguem and B Grabot ldquoGeneratingknowledge in maintenance from experience feedbackrdquoKnowledge-Based Systems vol 68 pp 4ndash20 2014

[12] S Moehrle and W Raskob ldquoStructuring and reusingknowledge from historical events for supporting nuclearemergency and remediation managementrdquo Engineering Ap-plications of Artificial Intelligence vol 46 pp 303ndash311 2015

[13] Q Meng Z Zhang X Wan and X Rong ldquoProperties ex-ploring and information mining in consumer communitynetwork a case of huawei pollen clubrdquo Complexity vol 2018Article ID 9470580 19 pages 2018

[14] R S Renu and GMocko ldquoComputing similarity of text-basedassembly processes for knowledge retrieval and reuserdquoJournal of Manufacturing Systems vol 39 pp 101ndash110 2016

[15] G E Modoni M Doukas W Terkaj M Sacco andD Mourtzis ldquoEnhancing factory data integration through thedevelopment of an ontology from the reference models reuse

to the semantic conversion of the legacy modelsrdquo In-ternational Journal of Computer Integrated Manufacturingvol 30 no 10 pp 1043ndash1059 2017

[16] E R Reyes S Negny G C Robles and J M Le LannldquoImprovement of online adaptation knowledge acquisitionand reuse in case-based reasoning application to processengineering designrdquo Engineering Applications of ArtificialIntelligence vol 41 pp 1ndash16 2015

[17] B Kamsu-Foguem and F H Abanda ldquoExperience modelingwith graphs encoded knowledge for construction industryrdquoComputers in Industry vol 70 pp 79ndash88 2015

[18] B Kamsu-Foguem F H Abanda M B Doumbouya andJ F Tchouanguem ldquoGraph-based ontology reasoning forformal verification of BREEAM rulesrdquo Cognitive SystemsResearch vol 55 pp 14ndash33 2019

[19] S H Othman and G Beydoun ldquoA metamodel-basedknowledge sharing system for disaster managementrdquo ExpertSystems with Applications vol 63 pp 49ndash65 2016

[20] D Xia X Lu H Li W Wang Y Li and Z Zhang ldquoAmapreduce-based parallel frequent pattern growth algorithmfor spatiotemporal association analysis of mobile trajectorybig datardquo Complexity vol 2018 Article ID 2818251 16 pages2018

[21] J Geng S Wang W Gan et al ldquoPromoting geospatial servicefrom information to knowledge with spatiotemporal se-manticsrdquo Complexity vol 2019 Article ID 9301420 14 pages2019

[22] A Nowak-Brzezinska ldquoEnhancing the efficiency of a decisionsupport system through the clustering of complex rule-basedknowledge bases and modification of the inference algo-rithmrdquo Complexity vol 2018 Article ID 2065491 14 pages2018

[23] E Maleki F Belkadi N Boli et al ldquoOntology-basedframework enabling smart product-service systems appli-cation of sensing systems for machine health monitoringrdquoIEEE Internet of ings Journal vol 5 no 6 pp 4496ndash45052018

[24] F H Abanda B Kamsu-Foguem and J H M TahldquoBIMmdashnew rules of measurement ontology for constructioncost estimationrdquo Engineering Science and Technology anInternational Journal vol 20 no 2 pp 443ndash459 2017

[25] B Kamsu-Foguem and P Tiako ldquoRisk information formal-isation with graphsrdquo Computers in Industry vol 85 pp 58ndash69 2017

[26] M Barzegar A Sadeghi-Niaraki and M Shakeri ldquoSpatialexperience based route finding using ontologiesrdquo ETRIJournal pp 1ndash11 2019

[27] E Camossi P Villa and L Mazzola ldquoSemantic-basedanomalous pattern discovery in moving object trajectoriesrdquo2013 httpsarxivorgabs13051946

[28] R Wannous J Malki A Bouju and C Vincent ldquoModellingmobile object activities based on trajectory ontology rulesconsidering spatial relationship rulesrdquo in Modeling Ap-proaches and Algorithms for Advanced Computer Applicationspp 249ndash258 Springer International Publishing BerlinGermany 2013

[29] Y Hu K Janowicz D Carral et al ldquoA geo-ontology designpattern for semantic trajectoriesrdquo in Proceedings of the In-ternational Conference on Spatial Information eorypp 438ndash456 Springer International Publishing ScarboroughUK September 2013

[30] M Baglioni J Macedo C Renso and M Wachowicz ldquoAnontology-based approach for the semantic modelling andreasoning on trajectoriesrdquo in Advances in Conceptual

14 Complexity

-e algorithm makes the route-finding systems smartand ubiquitous intended to act like experienced people Aubiquitous system is a smart system which acts not only likea person but also like an experienced human A person whoworks for some field of interest for many years gets a lot ofexperience-is person has much knowledge To transfer theknowledge of the experienced persons to the system we needthis algorithm-e proposed algorithm can receive and storethe knowledge of an experienced person and also accu-mulate the experiences of many experienced persons

31 Driversrsquo Experience Ontology In our ontology thedriversrsquo experience class includes two subclasses experi-mental routes and nonexperimental routes -e experi-mental route class encompasses the paths which the

experienced drivers have driven -ese routes are collectedvia an app and are converted to OWL classes -is classitself includes two subclasses traffic and without-trafficclasses -e paths which have been passed at peak traffictimes (7ndash10 AM and 16ndash20 PM) are categorized under thetraffic subclass -e nonexperimental route class encom-passes the paths obtained from the route-finding procedure-ese should also be converted to OWL classes In thedriversrsquo experience ontology for the sake of simplicity and inorder to improve usability only the names of routes aredefined as classes individuals of each class are not defined inthis file For each route class a separate OWL file is definedthe name of which is the same as the corresponding routeclass in the driversrsquo experience ontology -e constituentnodes of a route are stored as individuals in the corre-sponding OWL file using their identification (ID) code as

Any data Any place

Any time

Any network

Any service

Any device

Any user

Ubiquitousinformation

system

Others

Figure 1 Elements of ubiquitous computing

Data collection

Data preparation

Data modeling

Evaluation

Implementation

Monitoring

Ontology designcreating road networks

Predictiveanalytics

Determiningweights

Compare to Google maps based on route length and travel time

Adding new routesto ontology

Figure 2 Research methodology

Complexity 5

their name -e coordinates of each node are defined as dataproperties of individuals in the ontology

-e created ontology is shown in Figure 4 To save spaceonly a limited number of routes are shown here -is figuredepicts the ontology before the route-finding process Afterroute finding if a new path has been created and the timeentered by the user is within the pick range it will be storedin the traffic subclass of the nonexperimental route classOtherwise it will be stored in the nontraffic subclass Eachclass in the lowest level of ontology shows a route and itsname is defined based on its location and destination Forroutes which were traversed on nontraffic times the name ofclass would be location destination and for routes related totraffic classes the name of class would be location_desti-nation_t If the route was traversed on traffic times and it wasretrieved from algorithmrsquos results the name of class wouldbe location_destination_n_t -e flowchart of this ontologyis shown in Figure 5

32 Weighing Method In this paper a graph for a streetnetwork is constructed -e route finding is performed

based on this graph Each segment of the path is consideredas an edge of the graph -e weight of each edge is de-termined based on the number of times the drivers havepassed the corresponding segment Since this number mightbe unexpectedly great its normal form is computed asfollows

fn ei( 1113857 f ei( 1113857 minus fmin + 1fmax minus fmin + 1

(1)

where f(ei) is the number of passes of edge ei and fmin andfmax are the minimum and maximum number of the wholepasses in the graph For a segment which is not located inexperimental routes f(ei) 0 -erefore fmin 0 Gener-ally the weights of the edges of the street network arecomputed using the following equation

w ei( 1113857 maxfn(e)

Length (e)1113888 1113889 minus

fn ei( 1113857

Length ei( 1113857 (2)

where w(ei) denotes the weight of ith edge of the graphLength(ei) is the length of the edge and max(fn

(e)Length(e)) is the maximum number of passes of an edge

Evaluation

Calculating route length and travel time

Combining ontology with the cost model and

applying it to road network

Implementation ofroute-finding algorithm

Data collection

Computing frequency oftraversing a segment Data preprocessing

Determiningweighing method

Converting routes toOWL files

Cost model creation Ontology creation

Figure 3 Workflow of the proposed approach

6 Complexity

divided by the length of that edge Since edges with smallerweights are preferred in Dijkstrarsquos algorithm the values aresubtracted from the maximum value to inverse values

-e reason for choosing Dijkstrarsquos algorithm in thisresearch is that in the proposed algorithm in each exe-cution if a new path is created by an algorithm it will bestored in ontology in order to reuse it in future -ereforewe need to use an algorithm which provides an exact resultwith one hundred percent reliability because the results willbe reused in the future Among the shortest path algo-rithms breadth-first search (BFS) and depth-first search(DFS) can only be used in a weighted graph if the weightsare equal However in our graph weights which are thenumber of traverses by taxis are not equal Moreover theyare slower than in the Dijkstra algorithm [58] -e Greedyalgorithm uses heuristics and Alowast algorithms which are acombination of the Greedy algorithm and the Dijkstraalgorithm -erefore they provide an approximate resultnot an exact result

33 Ubiquitous Ontology-Based Route-Finding AlgorithmFigure 6 depicts the way our proposed algorithm worksFirst the user specifies the location the destination andthe time -en the algorithm determines whether asubclass with the specified location and destination existsin the ontology or not If the time chosen falls within therange of peak times the existing paths in the trafficsubclass are searched Otherwise the paths in the non-traffic subclass are examined When a path exists thestored route is displayed to the user and the algorithmterminates without needing to conduct route findingHowever when there is no corresponding path the al-gorithm determines whether the location and destinationexist in the experimental street network or not If bothexist in the network only the experimental network is usedto construct the graph and find the route (again peak timedetermines the subclass to choosemdashtraffic or nontraffic)

and the algorithm terminates If the location andor thedestination does not exist in the experimental streetnetwork the closest points of the network to them arefound -en a rectangle with a diameter (R2 R1 + 2radic2 h)greater than the distance between the location (or desti-nation) and the point is extracted from the main roadnetwork -e flowchart of the proposed hierarchical al-gorithm and the extraction of the rectangular region fromthe road network are shown in Figure 7 To clarify theroute-finding procedure consider a situation where boththe location and the destination do not exist in the ex-perimental network In such a situation the route findingincludes three stages (1) route finding from the location toits closest point in the experimental network (SSrsquo) (2)route finding from this point to the closest point of theexperimental network with respect to the destination(SrsquoDrsquo) and (3) route finding from the closest point of theexperimental network (with respect to the destination) tothe destination (DrsquoD) After route finding the resultingroute is stored in the driversrsquo experience ontology as aseparate OWL file Now requests with these locations anddestinations are responded to in real time without re-quiring additional route finding

4 Implementation

In this paper an ontology-based route-finding algorithmbased on the driversrsquo experience in ubiquitous GIS spacehas been proposed -e steps of implementing the algo-rithm are shown in Figure 8 In order to implement theproposed algorithm the required data have been collectedvia an app that stores taxi vehiclesrsquo routes throughOpenStreetMap (OSM) maps -ese routes and theircorresponding pick-up and drop-off stations are located inTehran Iran and are entered into the app by the drivers-en they are converted to shapefiles Preprocessing tasksare performed in the ArcGIS software package and theresults are stored in the Oracle database Each route is also

Figure 4 Driversrsquo experience ontology (to save space only a limited number of routes are shown here)

Complexity 7

converted to an OWL file using OWL Application Pro-gramming Interface (API) in Java Using these routes adriversrsquo experience ontology is separately constructed inProtege software A sample of a route stored in the app isshown in Figure 9

After data preparation the presented algorithm wasimplemented In each run of the app if a new route has beengenerated it will be stored as a class in the driversrsquo expe-rience ontology to be used in future requests without needfor additional processing As a result the routes are stored ina low-volume manner while data processing reduces witheven more runs of the algorithm Each route class in theontology occupies about 20ndash25 percent of its correspondingshapefile Moreover in the presented algorithm two roadnetworks existmdasha peak-time road network and a low-traffic

road network During peak times finding an optimum routebecomes important therefore the algorithm will utilizeDijkstrarsquos algorithm and present the optimum path to theuser

5 Evaluation

In order to evaluate the proposed method 10 different lo-cation-destination pairs covering most of Tehran city wereselected First the experimental road network was fullyconstructed (ER) -en it was constructed only consideringthe low-traffic experimental routes (ETR) Also the shortestpaths between the location-destinations were obtained usingDijkstrarsquos algorithm (DR) For all obtained routes the routelengths were calculated Using driversrsquo reports the travel

Collecting drivers routeswith app

Adding the name (location_destination) of

each new route as a subclass of nontraffic class of experimental

routes

Consider the segments of each path as individuals of

that classrsquo path

Considering the fields (X coordinates Y coordinates)

as the data property

Driversrsquo experienceontology

Experimental route classNonexperimental route class

Adding the name (location_destination) of

each new route as a subclass of nontraffic

class ofnonexperimental routes

Creating experimentaland nonexperimental

route classes

Storing each path in a separate OWL file as a class

Retrieving each class ifrequested by the user

Traffic class of experimental routes

Nontraffic class of experimental routes

Nontraffic class ofnonexperimental routes

Traffic class ofnonexperimental routes

Adding the name (location_destination_t)of each new route as a subclass of traffic class of experimental routes

Is the enteredtime by user is in range

of pick traffic times

Adding the name (location_destination_t)of each new route as a subclass of traffic class

of nonexperimental routes

No Yes

Figure 5 Flowchart of creating driversrsquo experience ontology

8 Complexity

times were also calculated Figure 10 illustrates the resultingroutes for a location-destination pair (Route 10) using thethree methods above

51 Evaluation Based on the Route Length Figure 11 shows adiagram of the length of the paths for the three methods-eroute lengths have been computed by accumulating their

Get the locationdestination and time

from user

Is anyroute with this locationand destination in the

nontraffic classes of driverrsquosexperience ontology

Showing the saved pathto the user

End

Are locationand destinationin experimentalroadnetwork

Using only experimentalroad network for route

finding

Using Dijkstrarsquosalgorithm for route

finding

Finding the nearest point of theexperimental road network to the location

and extraction of a rectangle from themain road network with diameter largerthan the distance between the location

and the nearest point in the experimentalroad network

Is location in experimental

road network

Is destination inexperimental

road network

Route finding with experimentalroad network and segments inthe rectangle(s) of main road

network

Storing route name as a subclass oftraffic class of nonexperimental

road network in driverrsquosexperiences ontology

Finding the nearest point of theexperimental road network to the

destination and extraction of a rectanglefrom the main road network with

diameter larger than the distance betweenthe destination and the nearest point in

the experimental road network

Storing route in aseparate OWL file with

its segments andproperties

Yes

Yes

No

NoNoYes

Is the entered

time in peaktraffic times

Is anyroute with this locationand destination in the

traffic classes of driverrsquosexperience ontology

Considering experimental roadnetwork according to trafficclass of experimental routes

Considering experimental roadnetwork according to nontraffic

class of experimental routes

No Yes

NoNo

Is entered time inpeak traffic times

Storing route name as a subclass ofnontraffic class of nonexperimental

road network indriverrsquos experiences ontology

Yes No

Yes

No

Figure 6 Proposed algorithm flowchart

Complexity 9

constituting segments Figure 12 depicts the ratio of routelengths obtained from the two proposed methods to that ofDijkstrarsquos algorithm Figure 13 compares the mean length ofthe 10 routes for each method

It is observable from the figures that route lengths in theshortest path algorithm are shorter than that of the other twomethods -ey have in contrast greater values when con-sidering high-traffic routes -e mean length of routes is1102 km for Dijkstrarsquos algorithm 1422 km when only con-sidering traffic routes and 134 km when considering all theexperimental (traffic and nontraffic) routes -is is due to taxidrivers who generally select the longest but simultaneously

fastest routes On the other hand Dijkstrarsquos algorithm doesnot consider travel time

52 Evaluation Based on Travel Time Figure 14 shows adiagram of the travel times for the three methods -etravel times have been obtained from different driverswho have driven these routes in peak times and in realconditions Figure 15 depicts the travel times obtainedfrom the two proposed methods to that of Dijkstrarsquos al-gorithm Figure 16 compares the mean travel times foreach method

Experimental road network layer

Basic road network layer

Drsquo

Drsquo

D

Srsquo

SrsquoS

(a)

S

R1

R2

h

D

(b)

Figure 7 (a) Hierarchical experimental road network (b) Extraction of rectangular region from main road network [39]

Trace preprocessing

Ontology design

Ontology-based route finding

Ubiquitous ontology-based route finding

User interface Storingroute if anew routeis created

OWL route filesUser Ontologicalroute-finding

algorithm

GIS and ontologyexperts

Driversrsquo experiences ontology OWL route files Ontology data

Taxiroutedata

Mobile appfor data

collection

Experiencedtaxi drivers

Examiningfrequency ofusing a road

segment

Costmodel

a7

149

10

2 116

15

b

c d

ef

Routenetwork

graph

Figure 8 Steps for implementing the proposed algorithm

10 Complexity

Figure 9 A sample of a stored route in the app which collects data

DR

ETR

ERLocationdestination

Final map

N

S

W E

Figure 10 Resulting routes for a location-destination pair (Route 10) using the three methods

DijkstraExperimental routesExperimental traffic routes

2 3 4 5 6 7 8 9 101Route number

02468

101214161820

Rout

e len

gth

(km

)

Figure 11 Comparison of route length for 10 evaluated routes of the three methods

Complexity 11

It is observable from the figures that travel timesobtained from the hierarchical experimental methodwhen only considering high-traffic conditions are lowesthowever the travel time of the shortest path is the highest-e mean travel time is 80 minutes for Dijkstrarsquos algo-rithm 61 minutes when only considering the high-trafficroutes and 775 minutes when considering all of theexperimental routes Although the routes are long whenconsidering high-traffic routes they are fast due to theselection of low-traffic routes in this case For examplewhen going to Tajrish Square from Valiasr SquareDijkstrarsquos algorithm suggests driving through ValiasrStreet which corresponds to the shortest but not fastestroute However the proposed ontology-based approachsuggests that the driver proceeds mainly along ModarresHighway which reduces the travel time by 35 minutes inhigh-traffic conditions

Another approach to evaluate the two proposedmethodsagainst Dijkstrarsquos algorithm is to find the ideal route amongthem In order to find the ideal route one can divide theroute length by its travel time Obviously when the routelengths for all routes or just two of them are equal the onewhich has the shortest travel time should be selected as theoptimum path On the other hand when the travel times areequal the one which has the shortest length should be se-lected as the optimum path Accordingly in eight cases (of10 cases) the ontology-based route finding considering onlyhigh-traffic routes (ETR) predominates In the other twocases the travel time was roughly equal however the routelengths obtained fromDijkstrarsquos algorithm had lower valuesTo determine the deviation of each method from the idealcase which is shown in Figure 17 the following formula isused

Fj travel length of method j

travel time of method j

j ETR ER Dikjestra

Fe ideal route fraction

Dj Fj minus Fe1113872 1113873

Fe

(3)

-erefore according to the evaluations above it can beclaimed that eliminating the experimental high-traffic routesin high-traffic conditions and utilizing the driversrsquo

02468

10121416

Mea

n of

rout

e len

gth

(km

)

Dijkstra Experimentalroutes

Method name

Experimentaltrafficroutes

Figure 13 Mean of route length for 3 methods

0

20

40

60

80

100

120

Trav

el ti

me (

min

)

1 2 3 4 5 6 7 8 9 10Route number

DijkstraExperimental routesExperimental traffic routes

Figure 14 Comparison of travel time for 10 evaluated routes of 3methods

002040608

112

1 2 3 4 5 6 7 8 9 10

ERDijkstraETRDijkstra

Figure 15 Comparison of travel time for the 2 modes of theproposed method with Dijkstrarsquos algorithm

Dijkstra020406080

Mea

n of

trav

el ti

me (

min

)

Experimentalroutes

Method name

Experimentaltrafficroutes

Figure 16 Mean of travel time for 3 methods

0

05

1

15

2

1 2 3 4 5 6 7 8 9 10

ERDijkstraETRDijkstra

Figure 12 Comparison of route length in 2 modes of the proposedalgorithm with Dijkstrarsquos algorithm

12 Complexity

experiences make the algorithm generate the optimum routethrough low-traffic parts of the network -is minimizestravel time which is the most important parameter in routefinding

6 Conclusions

Different people gain different spatial experiences Asthere are no systems to store these experiences theyevaporate over time-e present study provides the abilityto store spatial experiences and to make the existingsystems in ubiquitous GIS space intelligent In this paperan ontology-based route-finding algorithm in ubiquitousGIS space was designed and implemented using theroutes driven by the drivers of Tehran in high- and low-traffic conditions

In the proposed route-finding algorithm based on thedriversrsquo experiences ontology the number of times that eachsegment is passed by drivers is used to assign weights to thatsegment of the network First the user specifies the location thedestination and the time -en the algorithm determineswhether a subclass with the specified location and destinationexists in the ontology or not If the time chosen falls duringpeak times the existing paths in the traffic subclass are searchedthrough Otherwise the paths in the nontraffic subclass areexamined When a path exists the stored route is displayed tothe user and the algorithm terminates without needing toconduct route finding However when there is no corre-sponding path the algorithm determines whether the locationand destination exist in the experimental road network or notIf both exist in the network only the experimental is used toconstruct the graph and find the route (again peak time

determines the subclass to choosemdashtraffic or nontraffic) andthe algorithm terminates If the origin andor the destinationdoes not exist in the experimental road network the closestpoints of the network to them are found-en a rectangle witha diameter (R2R1 + 2radic2h) greater than the distance betweenthe location (or destination) and the point is extracted from themain road network -e main difference of this algorithm tothe previous experimental route-finding algorithms is the useof ontology and the construction of road networks onlyconsidering the low-traffic paths when there are high-trafficpaths in the network Since each newpath is stored as a separateOWL class the storage issues in the databases are resolved Inaddition the use of ontology causes the system to update everytime the algorithm runs and generates new paths Utilizing onlylow-traffic routes in high-traffic conditions allows the system tosuggest routes that are faster ensuring the optimum path basedon travel time is achieved According to our evaluations theproposed algorithm suggests routes that are longer but si-multaneously the fastest -is guarantees the appropriatenessof using such an algorithm in route finding In this researchany user can access the ontology from any location and at anytime to conduct route finding Moreover being intelligent theubiquitous environment can receive and store the experiences(any data element of ubiquitous GIS) of any person to utilizewithin its services in the future

More innovative applications of spatial experiencemodeling such as ambulance firefighting and tourismservices can be subjects for further considerations -eprocess of route finding for this category is also affected bytraffic as emergency services can also become stuck in trafficlike other drivers because there are no dedicated lines on allroads Due to the limitation in length of road and a largenumber of vehicles in the road in traffic times it is hard forother vehicles to provide a route for traversing of the am-bulance and they still need to find a way to avoid traffic-erefore the experiences from expert technicians in am-bulances or firefighters who have been in real event con-ditions can be modeled and used in the future In additionthis study considers only two factors distance and time forevaluation Other criteria including fuel consumption can beconsidered in future

Data Availability

-e authors are ready to share the research data on request

Disclosure

Maryam Barzegar and Abolghasem Sadeghi-Niaraki are tobe considered as co-first authors

Conflicts of Interest

-e authors declare that they have no conflicts of interest

Acknowledgments

-is research was supported by theMSIT (Ministry of Scienceand ICT) Korea under the ITRC (Information Technology

0

01

02

03

041

2

3

4

5

6

7

8

9

10

DijkstraExperimental routes

Experimental traffic routesIdeal

Figure 17 Comparison of routes from the 3methods with the idealroute

Complexity 13

Research Center) support program (IITP-2019-2016-0-00312)supervised by the IITP (Institute for Information andCommunications Technology Planning and Evaluation)

References

[1] I P Tussyadiah and F J Zach ldquo-e role of geo-basedtechnology in place experiencesrdquo Annals of Tourism Researchvol 39 no 2 pp 780ndash800 2012

[2] G M Honti and J Abonyi ldquoA review of semantic sensortechnologies in internet of things architecturesrdquo Complexityvol 2019 Article ID 6473160 21 pages 2019

[3] U H Govindarajan A J Trappey and C V TrappeyldquoImmersive technology for human-centric cyberphysicalsystems in complex manufacturing processes a compre-hensive overview of the global patent profile using collectiveintelligencerdquo Complexity vol 2018 Article ID 428363417 pages 2018

[4] B K Foguem T Coudert C Beler and L GenesteldquoKnowledge formalization in experience feedback processesan ontology-based approachrdquo Computers in Industry vol 59no 7 pp 694ndash710 2008

[5] N Lasierra F Roldan A Alesanco and J Garcıa ldquoTowardsimproving usage and management of supplies in healthcarean ontology-based solution for sharing knowledgerdquo ExpertSystems with Applications vol 41 no 14 pp 6261ndash6273 2014

[6] M-H Abel ldquoKnowledge map-based web platform to facilitateorganizational learning return of experiencesrdquo Computers inHuman Behavior vol 51 pp 960ndash966 2015

[7] A Garcıa-Crespo J L Lopez-Cuadrado R Colomo-PalaciosI Gonzalez-Carrasco and B Ruiz-Mezcua ldquoSem-fit a se-mantic based expert system to provide recommendations inthe tourism domainrdquo Expert Systems with Applicationsvol 38 no 10 pp 13310ndash13319 2011

[8] D Mourtzis M Doukas and C Giannoulis ldquoAn inference-based knowledge reuse framework for historical product andproduction information retrievalrdquo Procedia CIRP vol 41pp 472ndash477 2016

[9] K Efthymiou K Sipsas D Mourtzis and G ChryssolourisldquoOn knowledge reuse for manufacturing systems design andplanning a semantic technology approachrdquo CIRP Journal ofManufacturing Science and Technology vol 8 pp 1ndash11 2015

[10] W L Mikos J C E Ferreira P E A Botura and L S FreitasldquoA system for distributed sharing and reuse of design andmanufacturing knowledge in the PFMEA domain using adescription logics-based ontologyrdquo Journal of ManufacturingSystems vol 30 no 3 pp 133ndash143 2011

[11] P P Ruiz B K Foguem and B Grabot ldquoGeneratingknowledge in maintenance from experience feedbackrdquoKnowledge-Based Systems vol 68 pp 4ndash20 2014

[12] S Moehrle and W Raskob ldquoStructuring and reusingknowledge from historical events for supporting nuclearemergency and remediation managementrdquo Engineering Ap-plications of Artificial Intelligence vol 46 pp 303ndash311 2015

[13] Q Meng Z Zhang X Wan and X Rong ldquoProperties ex-ploring and information mining in consumer communitynetwork a case of huawei pollen clubrdquo Complexity vol 2018Article ID 9470580 19 pages 2018

[14] R S Renu and GMocko ldquoComputing similarity of text-basedassembly processes for knowledge retrieval and reuserdquoJournal of Manufacturing Systems vol 39 pp 101ndash110 2016

[15] G E Modoni M Doukas W Terkaj M Sacco andD Mourtzis ldquoEnhancing factory data integration through thedevelopment of an ontology from the reference models reuse

to the semantic conversion of the legacy modelsrdquo In-ternational Journal of Computer Integrated Manufacturingvol 30 no 10 pp 1043ndash1059 2017

[16] E R Reyes S Negny G C Robles and J M Le LannldquoImprovement of online adaptation knowledge acquisitionand reuse in case-based reasoning application to processengineering designrdquo Engineering Applications of ArtificialIntelligence vol 41 pp 1ndash16 2015

[17] B Kamsu-Foguem and F H Abanda ldquoExperience modelingwith graphs encoded knowledge for construction industryrdquoComputers in Industry vol 70 pp 79ndash88 2015

[18] B Kamsu-Foguem F H Abanda M B Doumbouya andJ F Tchouanguem ldquoGraph-based ontology reasoning forformal verification of BREEAM rulesrdquo Cognitive SystemsResearch vol 55 pp 14ndash33 2019

[19] S H Othman and G Beydoun ldquoA metamodel-basedknowledge sharing system for disaster managementrdquo ExpertSystems with Applications vol 63 pp 49ndash65 2016

[20] D Xia X Lu H Li W Wang Y Li and Z Zhang ldquoAmapreduce-based parallel frequent pattern growth algorithmfor spatiotemporal association analysis of mobile trajectorybig datardquo Complexity vol 2018 Article ID 2818251 16 pages2018

[21] J Geng S Wang W Gan et al ldquoPromoting geospatial servicefrom information to knowledge with spatiotemporal se-manticsrdquo Complexity vol 2019 Article ID 9301420 14 pages2019

[22] A Nowak-Brzezinska ldquoEnhancing the efficiency of a decisionsupport system through the clustering of complex rule-basedknowledge bases and modification of the inference algo-rithmrdquo Complexity vol 2018 Article ID 2065491 14 pages2018

[23] E Maleki F Belkadi N Boli et al ldquoOntology-basedframework enabling smart product-service systems appli-cation of sensing systems for machine health monitoringrdquoIEEE Internet of ings Journal vol 5 no 6 pp 4496ndash45052018

[24] F H Abanda B Kamsu-Foguem and J H M TahldquoBIMmdashnew rules of measurement ontology for constructioncost estimationrdquo Engineering Science and Technology anInternational Journal vol 20 no 2 pp 443ndash459 2017

[25] B Kamsu-Foguem and P Tiako ldquoRisk information formal-isation with graphsrdquo Computers in Industry vol 85 pp 58ndash69 2017

[26] M Barzegar A Sadeghi-Niaraki and M Shakeri ldquoSpatialexperience based route finding using ontologiesrdquo ETRIJournal pp 1ndash11 2019

[27] E Camossi P Villa and L Mazzola ldquoSemantic-basedanomalous pattern discovery in moving object trajectoriesrdquo2013 httpsarxivorgabs13051946

[28] R Wannous J Malki A Bouju and C Vincent ldquoModellingmobile object activities based on trajectory ontology rulesconsidering spatial relationship rulesrdquo in Modeling Ap-proaches and Algorithms for Advanced Computer Applicationspp 249ndash258 Springer International Publishing BerlinGermany 2013

[29] Y Hu K Janowicz D Carral et al ldquoA geo-ontology designpattern for semantic trajectoriesrdquo in Proceedings of the In-ternational Conference on Spatial Information eorypp 438ndash456 Springer International Publishing ScarboroughUK September 2013

[30] M Baglioni J Macedo C Renso and M Wachowicz ldquoAnontology-based approach for the semantic modelling andreasoning on trajectoriesrdquo in Advances in Conceptual

14 Complexity

their name -e coordinates of each node are defined as dataproperties of individuals in the ontology

-e created ontology is shown in Figure 4 To save spaceonly a limited number of routes are shown here -is figuredepicts the ontology before the route-finding process Afterroute finding if a new path has been created and the timeentered by the user is within the pick range it will be storedin the traffic subclass of the nonexperimental route classOtherwise it will be stored in the nontraffic subclass Eachclass in the lowest level of ontology shows a route and itsname is defined based on its location and destination Forroutes which were traversed on nontraffic times the name ofclass would be location destination and for routes related totraffic classes the name of class would be location_desti-nation_t If the route was traversed on traffic times and it wasretrieved from algorithmrsquos results the name of class wouldbe location_destination_n_t -e flowchart of this ontologyis shown in Figure 5

32 Weighing Method In this paper a graph for a streetnetwork is constructed -e route finding is performed

based on this graph Each segment of the path is consideredas an edge of the graph -e weight of each edge is de-termined based on the number of times the drivers havepassed the corresponding segment Since this number mightbe unexpectedly great its normal form is computed asfollows

fn ei( 1113857 f ei( 1113857 minus fmin + 1fmax minus fmin + 1

(1)

where f(ei) is the number of passes of edge ei and fmin andfmax are the minimum and maximum number of the wholepasses in the graph For a segment which is not located inexperimental routes f(ei) 0 -erefore fmin 0 Gener-ally the weights of the edges of the street network arecomputed using the following equation

w ei( 1113857 maxfn(e)

Length (e)1113888 1113889 minus

fn ei( 1113857

Length ei( 1113857 (2)

where w(ei) denotes the weight of ith edge of the graphLength(ei) is the length of the edge and max(fn

(e)Length(e)) is the maximum number of passes of an edge

Evaluation

Calculating route length and travel time

Combining ontology with the cost model and

applying it to road network

Implementation ofroute-finding algorithm

Data collection

Computing frequency oftraversing a segment Data preprocessing

Determiningweighing method

Converting routes toOWL files

Cost model creation Ontology creation

Figure 3 Workflow of the proposed approach

6 Complexity

divided by the length of that edge Since edges with smallerweights are preferred in Dijkstrarsquos algorithm the values aresubtracted from the maximum value to inverse values

-e reason for choosing Dijkstrarsquos algorithm in thisresearch is that in the proposed algorithm in each exe-cution if a new path is created by an algorithm it will bestored in ontology in order to reuse it in future -ereforewe need to use an algorithm which provides an exact resultwith one hundred percent reliability because the results willbe reused in the future Among the shortest path algo-rithms breadth-first search (BFS) and depth-first search(DFS) can only be used in a weighted graph if the weightsare equal However in our graph weights which are thenumber of traverses by taxis are not equal Moreover theyare slower than in the Dijkstra algorithm [58] -e Greedyalgorithm uses heuristics and Alowast algorithms which are acombination of the Greedy algorithm and the Dijkstraalgorithm -erefore they provide an approximate resultnot an exact result

33 Ubiquitous Ontology-Based Route-Finding AlgorithmFigure 6 depicts the way our proposed algorithm worksFirst the user specifies the location the destination andthe time -en the algorithm determines whether asubclass with the specified location and destination existsin the ontology or not If the time chosen falls within therange of peak times the existing paths in the trafficsubclass are searched Otherwise the paths in the non-traffic subclass are examined When a path exists thestored route is displayed to the user and the algorithmterminates without needing to conduct route findingHowever when there is no corresponding path the al-gorithm determines whether the location and destinationexist in the experimental street network or not If bothexist in the network only the experimental network is usedto construct the graph and find the route (again peak timedetermines the subclass to choosemdashtraffic or nontraffic)

and the algorithm terminates If the location andor thedestination does not exist in the experimental streetnetwork the closest points of the network to them arefound -en a rectangle with a diameter (R2 R1 + 2radic2 h)greater than the distance between the location (or desti-nation) and the point is extracted from the main roadnetwork -e flowchart of the proposed hierarchical al-gorithm and the extraction of the rectangular region fromthe road network are shown in Figure 7 To clarify theroute-finding procedure consider a situation where boththe location and the destination do not exist in the ex-perimental network In such a situation the route findingincludes three stages (1) route finding from the location toits closest point in the experimental network (SSrsquo) (2)route finding from this point to the closest point of theexperimental network with respect to the destination(SrsquoDrsquo) and (3) route finding from the closest point of theexperimental network (with respect to the destination) tothe destination (DrsquoD) After route finding the resultingroute is stored in the driversrsquo experience ontology as aseparate OWL file Now requests with these locations anddestinations are responded to in real time without re-quiring additional route finding

4 Implementation

In this paper an ontology-based route-finding algorithmbased on the driversrsquo experience in ubiquitous GIS spacehas been proposed -e steps of implementing the algo-rithm are shown in Figure 8 In order to implement theproposed algorithm the required data have been collectedvia an app that stores taxi vehiclesrsquo routes throughOpenStreetMap (OSM) maps -ese routes and theircorresponding pick-up and drop-off stations are located inTehran Iran and are entered into the app by the drivers-en they are converted to shapefiles Preprocessing tasksare performed in the ArcGIS software package and theresults are stored in the Oracle database Each route is also

Figure 4 Driversrsquo experience ontology (to save space only a limited number of routes are shown here)

Complexity 7

converted to an OWL file using OWL Application Pro-gramming Interface (API) in Java Using these routes adriversrsquo experience ontology is separately constructed inProtege software A sample of a route stored in the app isshown in Figure 9

After data preparation the presented algorithm wasimplemented In each run of the app if a new route has beengenerated it will be stored as a class in the driversrsquo expe-rience ontology to be used in future requests without needfor additional processing As a result the routes are stored ina low-volume manner while data processing reduces witheven more runs of the algorithm Each route class in theontology occupies about 20ndash25 percent of its correspondingshapefile Moreover in the presented algorithm two roadnetworks existmdasha peak-time road network and a low-traffic

road network During peak times finding an optimum routebecomes important therefore the algorithm will utilizeDijkstrarsquos algorithm and present the optimum path to theuser

5 Evaluation

In order to evaluate the proposed method 10 different lo-cation-destination pairs covering most of Tehran city wereselected First the experimental road network was fullyconstructed (ER) -en it was constructed only consideringthe low-traffic experimental routes (ETR) Also the shortestpaths between the location-destinations were obtained usingDijkstrarsquos algorithm (DR) For all obtained routes the routelengths were calculated Using driversrsquo reports the travel

Collecting drivers routeswith app

Adding the name (location_destination) of

each new route as a subclass of nontraffic class of experimental

routes

Consider the segments of each path as individuals of

that classrsquo path

Considering the fields (X coordinates Y coordinates)

as the data property

Driversrsquo experienceontology

Experimental route classNonexperimental route class

Adding the name (location_destination) of

each new route as a subclass of nontraffic

class ofnonexperimental routes

Creating experimentaland nonexperimental

route classes

Storing each path in a separate OWL file as a class

Retrieving each class ifrequested by the user

Traffic class of experimental routes

Nontraffic class of experimental routes

Nontraffic class ofnonexperimental routes

Traffic class ofnonexperimental routes

Adding the name (location_destination_t)of each new route as a subclass of traffic class of experimental routes

Is the enteredtime by user is in range

of pick traffic times

Adding the name (location_destination_t)of each new route as a subclass of traffic class

of nonexperimental routes

No Yes

Figure 5 Flowchart of creating driversrsquo experience ontology

8 Complexity

times were also calculated Figure 10 illustrates the resultingroutes for a location-destination pair (Route 10) using thethree methods above

51 Evaluation Based on the Route Length Figure 11 shows adiagram of the length of the paths for the three methods-eroute lengths have been computed by accumulating their

Get the locationdestination and time

from user

Is anyroute with this locationand destination in the

nontraffic classes of driverrsquosexperience ontology

Showing the saved pathto the user

End

Are locationand destinationin experimentalroadnetwork

Using only experimentalroad network for route

finding

Using Dijkstrarsquosalgorithm for route

finding

Finding the nearest point of theexperimental road network to the location

and extraction of a rectangle from themain road network with diameter largerthan the distance between the location

and the nearest point in the experimentalroad network

Is location in experimental

road network

Is destination inexperimental

road network

Route finding with experimentalroad network and segments inthe rectangle(s) of main road

network

Storing route name as a subclass oftraffic class of nonexperimental

road network in driverrsquosexperiences ontology

Finding the nearest point of theexperimental road network to the

destination and extraction of a rectanglefrom the main road network with

diameter larger than the distance betweenthe destination and the nearest point in

the experimental road network

Storing route in aseparate OWL file with

its segments andproperties

Yes

Yes

No

NoNoYes

Is the entered

time in peaktraffic times

Is anyroute with this locationand destination in the

traffic classes of driverrsquosexperience ontology

Considering experimental roadnetwork according to trafficclass of experimental routes

Considering experimental roadnetwork according to nontraffic

class of experimental routes

No Yes

NoNo

Is entered time inpeak traffic times

Storing route name as a subclass ofnontraffic class of nonexperimental

road network indriverrsquos experiences ontology

Yes No

Yes

No

Figure 6 Proposed algorithm flowchart

Complexity 9

constituting segments Figure 12 depicts the ratio of routelengths obtained from the two proposed methods to that ofDijkstrarsquos algorithm Figure 13 compares the mean length ofthe 10 routes for each method

It is observable from the figures that route lengths in theshortest path algorithm are shorter than that of the other twomethods -ey have in contrast greater values when con-sidering high-traffic routes -e mean length of routes is1102 km for Dijkstrarsquos algorithm 1422 km when only con-sidering traffic routes and 134 km when considering all theexperimental (traffic and nontraffic) routes -is is due to taxidrivers who generally select the longest but simultaneously

fastest routes On the other hand Dijkstrarsquos algorithm doesnot consider travel time

52 Evaluation Based on Travel Time Figure 14 shows adiagram of the travel times for the three methods -etravel times have been obtained from different driverswho have driven these routes in peak times and in realconditions Figure 15 depicts the travel times obtainedfrom the two proposed methods to that of Dijkstrarsquos al-gorithm Figure 16 compares the mean travel times foreach method

Experimental road network layer

Basic road network layer

Drsquo

Drsquo

D

Srsquo

SrsquoS

(a)

S

R1

R2

h

D

(b)

Figure 7 (a) Hierarchical experimental road network (b) Extraction of rectangular region from main road network [39]

Trace preprocessing

Ontology design

Ontology-based route finding

Ubiquitous ontology-based route finding

User interface Storingroute if anew routeis created

OWL route filesUser Ontologicalroute-finding

algorithm

GIS and ontologyexperts

Driversrsquo experiences ontology OWL route files Ontology data

Taxiroutedata

Mobile appfor data

collection

Experiencedtaxi drivers

Examiningfrequency ofusing a road

segment

Costmodel

a7

149

10

2 116

15

b

c d

ef

Routenetwork

graph

Figure 8 Steps for implementing the proposed algorithm

10 Complexity

Figure 9 A sample of a stored route in the app which collects data

DR

ETR

ERLocationdestination

Final map

N

S

W E

Figure 10 Resulting routes for a location-destination pair (Route 10) using the three methods

DijkstraExperimental routesExperimental traffic routes

2 3 4 5 6 7 8 9 101Route number

02468

101214161820

Rout

e len

gth

(km

)

Figure 11 Comparison of route length for 10 evaluated routes of the three methods

Complexity 11

It is observable from the figures that travel timesobtained from the hierarchical experimental methodwhen only considering high-traffic conditions are lowesthowever the travel time of the shortest path is the highest-e mean travel time is 80 minutes for Dijkstrarsquos algo-rithm 61 minutes when only considering the high-trafficroutes and 775 minutes when considering all of theexperimental routes Although the routes are long whenconsidering high-traffic routes they are fast due to theselection of low-traffic routes in this case For examplewhen going to Tajrish Square from Valiasr SquareDijkstrarsquos algorithm suggests driving through ValiasrStreet which corresponds to the shortest but not fastestroute However the proposed ontology-based approachsuggests that the driver proceeds mainly along ModarresHighway which reduces the travel time by 35 minutes inhigh-traffic conditions

Another approach to evaluate the two proposedmethodsagainst Dijkstrarsquos algorithm is to find the ideal route amongthem In order to find the ideal route one can divide theroute length by its travel time Obviously when the routelengths for all routes or just two of them are equal the onewhich has the shortest travel time should be selected as theoptimum path On the other hand when the travel times areequal the one which has the shortest length should be se-lected as the optimum path Accordingly in eight cases (of10 cases) the ontology-based route finding considering onlyhigh-traffic routes (ETR) predominates In the other twocases the travel time was roughly equal however the routelengths obtained fromDijkstrarsquos algorithm had lower valuesTo determine the deviation of each method from the idealcase which is shown in Figure 17 the following formula isused

Fj travel length of method j

travel time of method j

j ETR ER Dikjestra

Fe ideal route fraction

Dj Fj minus Fe1113872 1113873

Fe

(3)

-erefore according to the evaluations above it can beclaimed that eliminating the experimental high-traffic routesin high-traffic conditions and utilizing the driversrsquo

02468

10121416

Mea

n of

rout

e len

gth

(km

)

Dijkstra Experimentalroutes

Method name

Experimentaltrafficroutes

Figure 13 Mean of route length for 3 methods

0

20

40

60

80

100

120

Trav

el ti

me (

min

)

1 2 3 4 5 6 7 8 9 10Route number

DijkstraExperimental routesExperimental traffic routes

Figure 14 Comparison of travel time for 10 evaluated routes of 3methods

002040608

112

1 2 3 4 5 6 7 8 9 10

ERDijkstraETRDijkstra

Figure 15 Comparison of travel time for the 2 modes of theproposed method with Dijkstrarsquos algorithm

Dijkstra020406080

Mea

n of

trav

el ti

me (

min

)

Experimentalroutes

Method name

Experimentaltrafficroutes

Figure 16 Mean of travel time for 3 methods

0

05

1

15

2

1 2 3 4 5 6 7 8 9 10

ERDijkstraETRDijkstra

Figure 12 Comparison of route length in 2 modes of the proposedalgorithm with Dijkstrarsquos algorithm

12 Complexity

experiences make the algorithm generate the optimum routethrough low-traffic parts of the network -is minimizestravel time which is the most important parameter in routefinding

6 Conclusions

Different people gain different spatial experiences Asthere are no systems to store these experiences theyevaporate over time-e present study provides the abilityto store spatial experiences and to make the existingsystems in ubiquitous GIS space intelligent In this paperan ontology-based route-finding algorithm in ubiquitousGIS space was designed and implemented using theroutes driven by the drivers of Tehran in high- and low-traffic conditions

In the proposed route-finding algorithm based on thedriversrsquo experiences ontology the number of times that eachsegment is passed by drivers is used to assign weights to thatsegment of the network First the user specifies the location thedestination and the time -en the algorithm determineswhether a subclass with the specified location and destinationexists in the ontology or not If the time chosen falls duringpeak times the existing paths in the traffic subclass are searchedthrough Otherwise the paths in the nontraffic subclass areexamined When a path exists the stored route is displayed tothe user and the algorithm terminates without needing toconduct route finding However when there is no corre-sponding path the algorithm determines whether the locationand destination exist in the experimental road network or notIf both exist in the network only the experimental is used toconstruct the graph and find the route (again peak time

determines the subclass to choosemdashtraffic or nontraffic) andthe algorithm terminates If the origin andor the destinationdoes not exist in the experimental road network the closestpoints of the network to them are found-en a rectangle witha diameter (R2R1 + 2radic2h) greater than the distance betweenthe location (or destination) and the point is extracted from themain road network -e main difference of this algorithm tothe previous experimental route-finding algorithms is the useof ontology and the construction of road networks onlyconsidering the low-traffic paths when there are high-trafficpaths in the network Since each newpath is stored as a separateOWL class the storage issues in the databases are resolved Inaddition the use of ontology causes the system to update everytime the algorithm runs and generates new paths Utilizing onlylow-traffic routes in high-traffic conditions allows the system tosuggest routes that are faster ensuring the optimum path basedon travel time is achieved According to our evaluations theproposed algorithm suggests routes that are longer but si-multaneously the fastest -is guarantees the appropriatenessof using such an algorithm in route finding In this researchany user can access the ontology from any location and at anytime to conduct route finding Moreover being intelligent theubiquitous environment can receive and store the experiences(any data element of ubiquitous GIS) of any person to utilizewithin its services in the future

More innovative applications of spatial experiencemodeling such as ambulance firefighting and tourismservices can be subjects for further considerations -eprocess of route finding for this category is also affected bytraffic as emergency services can also become stuck in trafficlike other drivers because there are no dedicated lines on allroads Due to the limitation in length of road and a largenumber of vehicles in the road in traffic times it is hard forother vehicles to provide a route for traversing of the am-bulance and they still need to find a way to avoid traffic-erefore the experiences from expert technicians in am-bulances or firefighters who have been in real event con-ditions can be modeled and used in the future In additionthis study considers only two factors distance and time forevaluation Other criteria including fuel consumption can beconsidered in future

Data Availability

-e authors are ready to share the research data on request

Disclosure

Maryam Barzegar and Abolghasem Sadeghi-Niaraki are tobe considered as co-first authors

Conflicts of Interest

-e authors declare that they have no conflicts of interest

Acknowledgments

-is research was supported by theMSIT (Ministry of Scienceand ICT) Korea under the ITRC (Information Technology

0

01

02

03

041

2

3

4

5

6

7

8

9

10

DijkstraExperimental routes

Experimental traffic routesIdeal

Figure 17 Comparison of routes from the 3methods with the idealroute

Complexity 13

Research Center) support program (IITP-2019-2016-0-00312)supervised by the IITP (Institute for Information andCommunications Technology Planning and Evaluation)

References

[1] I P Tussyadiah and F J Zach ldquo-e role of geo-basedtechnology in place experiencesrdquo Annals of Tourism Researchvol 39 no 2 pp 780ndash800 2012

[2] G M Honti and J Abonyi ldquoA review of semantic sensortechnologies in internet of things architecturesrdquo Complexityvol 2019 Article ID 6473160 21 pages 2019

[3] U H Govindarajan A J Trappey and C V TrappeyldquoImmersive technology for human-centric cyberphysicalsystems in complex manufacturing processes a compre-hensive overview of the global patent profile using collectiveintelligencerdquo Complexity vol 2018 Article ID 428363417 pages 2018

[4] B K Foguem T Coudert C Beler and L GenesteldquoKnowledge formalization in experience feedback processesan ontology-based approachrdquo Computers in Industry vol 59no 7 pp 694ndash710 2008

[5] N Lasierra F Roldan A Alesanco and J Garcıa ldquoTowardsimproving usage and management of supplies in healthcarean ontology-based solution for sharing knowledgerdquo ExpertSystems with Applications vol 41 no 14 pp 6261ndash6273 2014

[6] M-H Abel ldquoKnowledge map-based web platform to facilitateorganizational learning return of experiencesrdquo Computers inHuman Behavior vol 51 pp 960ndash966 2015

[7] A Garcıa-Crespo J L Lopez-Cuadrado R Colomo-PalaciosI Gonzalez-Carrasco and B Ruiz-Mezcua ldquoSem-fit a se-mantic based expert system to provide recommendations inthe tourism domainrdquo Expert Systems with Applicationsvol 38 no 10 pp 13310ndash13319 2011

[8] D Mourtzis M Doukas and C Giannoulis ldquoAn inference-based knowledge reuse framework for historical product andproduction information retrievalrdquo Procedia CIRP vol 41pp 472ndash477 2016

[9] K Efthymiou K Sipsas D Mourtzis and G ChryssolourisldquoOn knowledge reuse for manufacturing systems design andplanning a semantic technology approachrdquo CIRP Journal ofManufacturing Science and Technology vol 8 pp 1ndash11 2015

[10] W L Mikos J C E Ferreira P E A Botura and L S FreitasldquoA system for distributed sharing and reuse of design andmanufacturing knowledge in the PFMEA domain using adescription logics-based ontologyrdquo Journal of ManufacturingSystems vol 30 no 3 pp 133ndash143 2011

[11] P P Ruiz B K Foguem and B Grabot ldquoGeneratingknowledge in maintenance from experience feedbackrdquoKnowledge-Based Systems vol 68 pp 4ndash20 2014

[12] S Moehrle and W Raskob ldquoStructuring and reusingknowledge from historical events for supporting nuclearemergency and remediation managementrdquo Engineering Ap-plications of Artificial Intelligence vol 46 pp 303ndash311 2015

[13] Q Meng Z Zhang X Wan and X Rong ldquoProperties ex-ploring and information mining in consumer communitynetwork a case of huawei pollen clubrdquo Complexity vol 2018Article ID 9470580 19 pages 2018

[14] R S Renu and GMocko ldquoComputing similarity of text-basedassembly processes for knowledge retrieval and reuserdquoJournal of Manufacturing Systems vol 39 pp 101ndash110 2016

[15] G E Modoni M Doukas W Terkaj M Sacco andD Mourtzis ldquoEnhancing factory data integration through thedevelopment of an ontology from the reference models reuse

to the semantic conversion of the legacy modelsrdquo In-ternational Journal of Computer Integrated Manufacturingvol 30 no 10 pp 1043ndash1059 2017

[16] E R Reyes S Negny G C Robles and J M Le LannldquoImprovement of online adaptation knowledge acquisitionand reuse in case-based reasoning application to processengineering designrdquo Engineering Applications of ArtificialIntelligence vol 41 pp 1ndash16 2015

[17] B Kamsu-Foguem and F H Abanda ldquoExperience modelingwith graphs encoded knowledge for construction industryrdquoComputers in Industry vol 70 pp 79ndash88 2015

[18] B Kamsu-Foguem F H Abanda M B Doumbouya andJ F Tchouanguem ldquoGraph-based ontology reasoning forformal verification of BREEAM rulesrdquo Cognitive SystemsResearch vol 55 pp 14ndash33 2019

[19] S H Othman and G Beydoun ldquoA metamodel-basedknowledge sharing system for disaster managementrdquo ExpertSystems with Applications vol 63 pp 49ndash65 2016

[20] D Xia X Lu H Li W Wang Y Li and Z Zhang ldquoAmapreduce-based parallel frequent pattern growth algorithmfor spatiotemporal association analysis of mobile trajectorybig datardquo Complexity vol 2018 Article ID 2818251 16 pages2018

[21] J Geng S Wang W Gan et al ldquoPromoting geospatial servicefrom information to knowledge with spatiotemporal se-manticsrdquo Complexity vol 2019 Article ID 9301420 14 pages2019

[22] A Nowak-Brzezinska ldquoEnhancing the efficiency of a decisionsupport system through the clustering of complex rule-basedknowledge bases and modification of the inference algo-rithmrdquo Complexity vol 2018 Article ID 2065491 14 pages2018

[23] E Maleki F Belkadi N Boli et al ldquoOntology-basedframework enabling smart product-service systems appli-cation of sensing systems for machine health monitoringrdquoIEEE Internet of ings Journal vol 5 no 6 pp 4496ndash45052018

[24] F H Abanda B Kamsu-Foguem and J H M TahldquoBIMmdashnew rules of measurement ontology for constructioncost estimationrdquo Engineering Science and Technology anInternational Journal vol 20 no 2 pp 443ndash459 2017

[25] B Kamsu-Foguem and P Tiako ldquoRisk information formal-isation with graphsrdquo Computers in Industry vol 85 pp 58ndash69 2017

[26] M Barzegar A Sadeghi-Niaraki and M Shakeri ldquoSpatialexperience based route finding using ontologiesrdquo ETRIJournal pp 1ndash11 2019

[27] E Camossi P Villa and L Mazzola ldquoSemantic-basedanomalous pattern discovery in moving object trajectoriesrdquo2013 httpsarxivorgabs13051946

[28] R Wannous J Malki A Bouju and C Vincent ldquoModellingmobile object activities based on trajectory ontology rulesconsidering spatial relationship rulesrdquo in Modeling Ap-proaches and Algorithms for Advanced Computer Applicationspp 249ndash258 Springer International Publishing BerlinGermany 2013

[29] Y Hu K Janowicz D Carral et al ldquoA geo-ontology designpattern for semantic trajectoriesrdquo in Proceedings of the In-ternational Conference on Spatial Information eorypp 438ndash456 Springer International Publishing ScarboroughUK September 2013

[30] M Baglioni J Macedo C Renso and M Wachowicz ldquoAnontology-based approach for the semantic modelling andreasoning on trajectoriesrdquo in Advances in Conceptual

14 Complexity

divided by the length of that edge Since edges with smallerweights are preferred in Dijkstrarsquos algorithm the values aresubtracted from the maximum value to inverse values

-e reason for choosing Dijkstrarsquos algorithm in thisresearch is that in the proposed algorithm in each exe-cution if a new path is created by an algorithm it will bestored in ontology in order to reuse it in future -ereforewe need to use an algorithm which provides an exact resultwith one hundred percent reliability because the results willbe reused in the future Among the shortest path algo-rithms breadth-first search (BFS) and depth-first search(DFS) can only be used in a weighted graph if the weightsare equal However in our graph weights which are thenumber of traverses by taxis are not equal Moreover theyare slower than in the Dijkstra algorithm [58] -e Greedyalgorithm uses heuristics and Alowast algorithms which are acombination of the Greedy algorithm and the Dijkstraalgorithm -erefore they provide an approximate resultnot an exact result

33 Ubiquitous Ontology-Based Route-Finding AlgorithmFigure 6 depicts the way our proposed algorithm worksFirst the user specifies the location the destination andthe time -en the algorithm determines whether asubclass with the specified location and destination existsin the ontology or not If the time chosen falls within therange of peak times the existing paths in the trafficsubclass are searched Otherwise the paths in the non-traffic subclass are examined When a path exists thestored route is displayed to the user and the algorithmterminates without needing to conduct route findingHowever when there is no corresponding path the al-gorithm determines whether the location and destinationexist in the experimental street network or not If bothexist in the network only the experimental network is usedto construct the graph and find the route (again peak timedetermines the subclass to choosemdashtraffic or nontraffic)

and the algorithm terminates If the location andor thedestination does not exist in the experimental streetnetwork the closest points of the network to them arefound -en a rectangle with a diameter (R2 R1 + 2radic2 h)greater than the distance between the location (or desti-nation) and the point is extracted from the main roadnetwork -e flowchart of the proposed hierarchical al-gorithm and the extraction of the rectangular region fromthe road network are shown in Figure 7 To clarify theroute-finding procedure consider a situation where boththe location and the destination do not exist in the ex-perimental network In such a situation the route findingincludes three stages (1) route finding from the location toits closest point in the experimental network (SSrsquo) (2)route finding from this point to the closest point of theexperimental network with respect to the destination(SrsquoDrsquo) and (3) route finding from the closest point of theexperimental network (with respect to the destination) tothe destination (DrsquoD) After route finding the resultingroute is stored in the driversrsquo experience ontology as aseparate OWL file Now requests with these locations anddestinations are responded to in real time without re-quiring additional route finding

4 Implementation

In this paper an ontology-based route-finding algorithmbased on the driversrsquo experience in ubiquitous GIS spacehas been proposed -e steps of implementing the algo-rithm are shown in Figure 8 In order to implement theproposed algorithm the required data have been collectedvia an app that stores taxi vehiclesrsquo routes throughOpenStreetMap (OSM) maps -ese routes and theircorresponding pick-up and drop-off stations are located inTehran Iran and are entered into the app by the drivers-en they are converted to shapefiles Preprocessing tasksare performed in the ArcGIS software package and theresults are stored in the Oracle database Each route is also

Figure 4 Driversrsquo experience ontology (to save space only a limited number of routes are shown here)

Complexity 7

converted to an OWL file using OWL Application Pro-gramming Interface (API) in Java Using these routes adriversrsquo experience ontology is separately constructed inProtege software A sample of a route stored in the app isshown in Figure 9

After data preparation the presented algorithm wasimplemented In each run of the app if a new route has beengenerated it will be stored as a class in the driversrsquo expe-rience ontology to be used in future requests without needfor additional processing As a result the routes are stored ina low-volume manner while data processing reduces witheven more runs of the algorithm Each route class in theontology occupies about 20ndash25 percent of its correspondingshapefile Moreover in the presented algorithm two roadnetworks existmdasha peak-time road network and a low-traffic

road network During peak times finding an optimum routebecomes important therefore the algorithm will utilizeDijkstrarsquos algorithm and present the optimum path to theuser

5 Evaluation

In order to evaluate the proposed method 10 different lo-cation-destination pairs covering most of Tehran city wereselected First the experimental road network was fullyconstructed (ER) -en it was constructed only consideringthe low-traffic experimental routes (ETR) Also the shortestpaths between the location-destinations were obtained usingDijkstrarsquos algorithm (DR) For all obtained routes the routelengths were calculated Using driversrsquo reports the travel

Collecting drivers routeswith app

Adding the name (location_destination) of

each new route as a subclass of nontraffic class of experimental

routes

Consider the segments of each path as individuals of

that classrsquo path

Considering the fields (X coordinates Y coordinates)

as the data property

Driversrsquo experienceontology

Experimental route classNonexperimental route class

Adding the name (location_destination) of

each new route as a subclass of nontraffic

class ofnonexperimental routes

Creating experimentaland nonexperimental

route classes

Storing each path in a separate OWL file as a class

Retrieving each class ifrequested by the user

Traffic class of experimental routes

Nontraffic class of experimental routes

Nontraffic class ofnonexperimental routes

Traffic class ofnonexperimental routes

Adding the name (location_destination_t)of each new route as a subclass of traffic class of experimental routes

Is the enteredtime by user is in range

of pick traffic times

Adding the name (location_destination_t)of each new route as a subclass of traffic class

of nonexperimental routes

No Yes

Figure 5 Flowchart of creating driversrsquo experience ontology

8 Complexity

times were also calculated Figure 10 illustrates the resultingroutes for a location-destination pair (Route 10) using thethree methods above

51 Evaluation Based on the Route Length Figure 11 shows adiagram of the length of the paths for the three methods-eroute lengths have been computed by accumulating their

Get the locationdestination and time

from user

Is anyroute with this locationand destination in the

nontraffic classes of driverrsquosexperience ontology

Showing the saved pathto the user

End

Are locationand destinationin experimentalroadnetwork

Using only experimentalroad network for route

finding

Using Dijkstrarsquosalgorithm for route

finding

Finding the nearest point of theexperimental road network to the location

and extraction of a rectangle from themain road network with diameter largerthan the distance between the location

and the nearest point in the experimentalroad network

Is location in experimental

road network

Is destination inexperimental

road network

Route finding with experimentalroad network and segments inthe rectangle(s) of main road

network

Storing route name as a subclass oftraffic class of nonexperimental

road network in driverrsquosexperiences ontology

Finding the nearest point of theexperimental road network to the

destination and extraction of a rectanglefrom the main road network with

diameter larger than the distance betweenthe destination and the nearest point in

the experimental road network

Storing route in aseparate OWL file with

its segments andproperties

Yes

Yes

No

NoNoYes

Is the entered

time in peaktraffic times

Is anyroute with this locationand destination in the

traffic classes of driverrsquosexperience ontology

Considering experimental roadnetwork according to trafficclass of experimental routes

Considering experimental roadnetwork according to nontraffic

class of experimental routes

No Yes

NoNo

Is entered time inpeak traffic times

Storing route name as a subclass ofnontraffic class of nonexperimental

road network indriverrsquos experiences ontology

Yes No

Yes

No

Figure 6 Proposed algorithm flowchart

Complexity 9

constituting segments Figure 12 depicts the ratio of routelengths obtained from the two proposed methods to that ofDijkstrarsquos algorithm Figure 13 compares the mean length ofthe 10 routes for each method

It is observable from the figures that route lengths in theshortest path algorithm are shorter than that of the other twomethods -ey have in contrast greater values when con-sidering high-traffic routes -e mean length of routes is1102 km for Dijkstrarsquos algorithm 1422 km when only con-sidering traffic routes and 134 km when considering all theexperimental (traffic and nontraffic) routes -is is due to taxidrivers who generally select the longest but simultaneously

fastest routes On the other hand Dijkstrarsquos algorithm doesnot consider travel time

52 Evaluation Based on Travel Time Figure 14 shows adiagram of the travel times for the three methods -etravel times have been obtained from different driverswho have driven these routes in peak times and in realconditions Figure 15 depicts the travel times obtainedfrom the two proposed methods to that of Dijkstrarsquos al-gorithm Figure 16 compares the mean travel times foreach method

Experimental road network layer

Basic road network layer

Drsquo

Drsquo

D

Srsquo

SrsquoS

(a)

S

R1

R2

h

D

(b)

Figure 7 (a) Hierarchical experimental road network (b) Extraction of rectangular region from main road network [39]

Trace preprocessing

Ontology design

Ontology-based route finding

Ubiquitous ontology-based route finding

User interface Storingroute if anew routeis created

OWL route filesUser Ontologicalroute-finding

algorithm

GIS and ontologyexperts

Driversrsquo experiences ontology OWL route files Ontology data

Taxiroutedata

Mobile appfor data

collection

Experiencedtaxi drivers

Examiningfrequency ofusing a road

segment

Costmodel

a7

149

10

2 116

15

b

c d

ef

Routenetwork

graph

Figure 8 Steps for implementing the proposed algorithm

10 Complexity

Figure 9 A sample of a stored route in the app which collects data

DR

ETR

ERLocationdestination

Final map

N

S

W E

Figure 10 Resulting routes for a location-destination pair (Route 10) using the three methods

DijkstraExperimental routesExperimental traffic routes

2 3 4 5 6 7 8 9 101Route number

02468

101214161820

Rout

e len

gth

(km

)

Figure 11 Comparison of route length for 10 evaluated routes of the three methods

Complexity 11

It is observable from the figures that travel timesobtained from the hierarchical experimental methodwhen only considering high-traffic conditions are lowesthowever the travel time of the shortest path is the highest-e mean travel time is 80 minutes for Dijkstrarsquos algo-rithm 61 minutes when only considering the high-trafficroutes and 775 minutes when considering all of theexperimental routes Although the routes are long whenconsidering high-traffic routes they are fast due to theselection of low-traffic routes in this case For examplewhen going to Tajrish Square from Valiasr SquareDijkstrarsquos algorithm suggests driving through ValiasrStreet which corresponds to the shortest but not fastestroute However the proposed ontology-based approachsuggests that the driver proceeds mainly along ModarresHighway which reduces the travel time by 35 minutes inhigh-traffic conditions

Another approach to evaluate the two proposedmethodsagainst Dijkstrarsquos algorithm is to find the ideal route amongthem In order to find the ideal route one can divide theroute length by its travel time Obviously when the routelengths for all routes or just two of them are equal the onewhich has the shortest travel time should be selected as theoptimum path On the other hand when the travel times areequal the one which has the shortest length should be se-lected as the optimum path Accordingly in eight cases (of10 cases) the ontology-based route finding considering onlyhigh-traffic routes (ETR) predominates In the other twocases the travel time was roughly equal however the routelengths obtained fromDijkstrarsquos algorithm had lower valuesTo determine the deviation of each method from the idealcase which is shown in Figure 17 the following formula isused

Fj travel length of method j

travel time of method j

j ETR ER Dikjestra

Fe ideal route fraction

Dj Fj minus Fe1113872 1113873

Fe

(3)

-erefore according to the evaluations above it can beclaimed that eliminating the experimental high-traffic routesin high-traffic conditions and utilizing the driversrsquo

02468

10121416

Mea

n of

rout

e len

gth

(km

)

Dijkstra Experimentalroutes

Method name

Experimentaltrafficroutes

Figure 13 Mean of route length for 3 methods

0

20

40

60

80

100

120

Trav

el ti

me (

min

)

1 2 3 4 5 6 7 8 9 10Route number

DijkstraExperimental routesExperimental traffic routes

Figure 14 Comparison of travel time for 10 evaluated routes of 3methods

002040608

112

1 2 3 4 5 6 7 8 9 10

ERDijkstraETRDijkstra

Figure 15 Comparison of travel time for the 2 modes of theproposed method with Dijkstrarsquos algorithm

Dijkstra020406080

Mea

n of

trav

el ti

me (

min

)

Experimentalroutes

Method name

Experimentaltrafficroutes

Figure 16 Mean of travel time for 3 methods

0

05

1

15

2

1 2 3 4 5 6 7 8 9 10

ERDijkstraETRDijkstra

Figure 12 Comparison of route length in 2 modes of the proposedalgorithm with Dijkstrarsquos algorithm

12 Complexity

experiences make the algorithm generate the optimum routethrough low-traffic parts of the network -is minimizestravel time which is the most important parameter in routefinding

6 Conclusions

Different people gain different spatial experiences Asthere are no systems to store these experiences theyevaporate over time-e present study provides the abilityto store spatial experiences and to make the existingsystems in ubiquitous GIS space intelligent In this paperan ontology-based route-finding algorithm in ubiquitousGIS space was designed and implemented using theroutes driven by the drivers of Tehran in high- and low-traffic conditions

In the proposed route-finding algorithm based on thedriversrsquo experiences ontology the number of times that eachsegment is passed by drivers is used to assign weights to thatsegment of the network First the user specifies the location thedestination and the time -en the algorithm determineswhether a subclass with the specified location and destinationexists in the ontology or not If the time chosen falls duringpeak times the existing paths in the traffic subclass are searchedthrough Otherwise the paths in the nontraffic subclass areexamined When a path exists the stored route is displayed tothe user and the algorithm terminates without needing toconduct route finding However when there is no corre-sponding path the algorithm determines whether the locationand destination exist in the experimental road network or notIf both exist in the network only the experimental is used toconstruct the graph and find the route (again peak time

determines the subclass to choosemdashtraffic or nontraffic) andthe algorithm terminates If the origin andor the destinationdoes not exist in the experimental road network the closestpoints of the network to them are found-en a rectangle witha diameter (R2R1 + 2radic2h) greater than the distance betweenthe location (or destination) and the point is extracted from themain road network -e main difference of this algorithm tothe previous experimental route-finding algorithms is the useof ontology and the construction of road networks onlyconsidering the low-traffic paths when there are high-trafficpaths in the network Since each newpath is stored as a separateOWL class the storage issues in the databases are resolved Inaddition the use of ontology causes the system to update everytime the algorithm runs and generates new paths Utilizing onlylow-traffic routes in high-traffic conditions allows the system tosuggest routes that are faster ensuring the optimum path basedon travel time is achieved According to our evaluations theproposed algorithm suggests routes that are longer but si-multaneously the fastest -is guarantees the appropriatenessof using such an algorithm in route finding In this researchany user can access the ontology from any location and at anytime to conduct route finding Moreover being intelligent theubiquitous environment can receive and store the experiences(any data element of ubiquitous GIS) of any person to utilizewithin its services in the future

More innovative applications of spatial experiencemodeling such as ambulance firefighting and tourismservices can be subjects for further considerations -eprocess of route finding for this category is also affected bytraffic as emergency services can also become stuck in trafficlike other drivers because there are no dedicated lines on allroads Due to the limitation in length of road and a largenumber of vehicles in the road in traffic times it is hard forother vehicles to provide a route for traversing of the am-bulance and they still need to find a way to avoid traffic-erefore the experiences from expert technicians in am-bulances or firefighters who have been in real event con-ditions can be modeled and used in the future In additionthis study considers only two factors distance and time forevaluation Other criteria including fuel consumption can beconsidered in future

Data Availability

-e authors are ready to share the research data on request

Disclosure

Maryam Barzegar and Abolghasem Sadeghi-Niaraki are tobe considered as co-first authors

Conflicts of Interest

-e authors declare that they have no conflicts of interest

Acknowledgments

-is research was supported by theMSIT (Ministry of Scienceand ICT) Korea under the ITRC (Information Technology

0

01

02

03

041

2

3

4

5

6

7

8

9

10

DijkstraExperimental routes

Experimental traffic routesIdeal

Figure 17 Comparison of routes from the 3methods with the idealroute

Complexity 13

Research Center) support program (IITP-2019-2016-0-00312)supervised by the IITP (Institute for Information andCommunications Technology Planning and Evaluation)

References

[1] I P Tussyadiah and F J Zach ldquo-e role of geo-basedtechnology in place experiencesrdquo Annals of Tourism Researchvol 39 no 2 pp 780ndash800 2012

[2] G M Honti and J Abonyi ldquoA review of semantic sensortechnologies in internet of things architecturesrdquo Complexityvol 2019 Article ID 6473160 21 pages 2019

[3] U H Govindarajan A J Trappey and C V TrappeyldquoImmersive technology for human-centric cyberphysicalsystems in complex manufacturing processes a compre-hensive overview of the global patent profile using collectiveintelligencerdquo Complexity vol 2018 Article ID 428363417 pages 2018

[4] B K Foguem T Coudert C Beler and L GenesteldquoKnowledge formalization in experience feedback processesan ontology-based approachrdquo Computers in Industry vol 59no 7 pp 694ndash710 2008

[5] N Lasierra F Roldan A Alesanco and J Garcıa ldquoTowardsimproving usage and management of supplies in healthcarean ontology-based solution for sharing knowledgerdquo ExpertSystems with Applications vol 41 no 14 pp 6261ndash6273 2014

[6] M-H Abel ldquoKnowledge map-based web platform to facilitateorganizational learning return of experiencesrdquo Computers inHuman Behavior vol 51 pp 960ndash966 2015

[7] A Garcıa-Crespo J L Lopez-Cuadrado R Colomo-PalaciosI Gonzalez-Carrasco and B Ruiz-Mezcua ldquoSem-fit a se-mantic based expert system to provide recommendations inthe tourism domainrdquo Expert Systems with Applicationsvol 38 no 10 pp 13310ndash13319 2011

[8] D Mourtzis M Doukas and C Giannoulis ldquoAn inference-based knowledge reuse framework for historical product andproduction information retrievalrdquo Procedia CIRP vol 41pp 472ndash477 2016

[9] K Efthymiou K Sipsas D Mourtzis and G ChryssolourisldquoOn knowledge reuse for manufacturing systems design andplanning a semantic technology approachrdquo CIRP Journal ofManufacturing Science and Technology vol 8 pp 1ndash11 2015

[10] W L Mikos J C E Ferreira P E A Botura and L S FreitasldquoA system for distributed sharing and reuse of design andmanufacturing knowledge in the PFMEA domain using adescription logics-based ontologyrdquo Journal of ManufacturingSystems vol 30 no 3 pp 133ndash143 2011

[11] P P Ruiz B K Foguem and B Grabot ldquoGeneratingknowledge in maintenance from experience feedbackrdquoKnowledge-Based Systems vol 68 pp 4ndash20 2014

[12] S Moehrle and W Raskob ldquoStructuring and reusingknowledge from historical events for supporting nuclearemergency and remediation managementrdquo Engineering Ap-plications of Artificial Intelligence vol 46 pp 303ndash311 2015

[13] Q Meng Z Zhang X Wan and X Rong ldquoProperties ex-ploring and information mining in consumer communitynetwork a case of huawei pollen clubrdquo Complexity vol 2018Article ID 9470580 19 pages 2018

[14] R S Renu and GMocko ldquoComputing similarity of text-basedassembly processes for knowledge retrieval and reuserdquoJournal of Manufacturing Systems vol 39 pp 101ndash110 2016

[15] G E Modoni M Doukas W Terkaj M Sacco andD Mourtzis ldquoEnhancing factory data integration through thedevelopment of an ontology from the reference models reuse

to the semantic conversion of the legacy modelsrdquo In-ternational Journal of Computer Integrated Manufacturingvol 30 no 10 pp 1043ndash1059 2017

[16] E R Reyes S Negny G C Robles and J M Le LannldquoImprovement of online adaptation knowledge acquisitionand reuse in case-based reasoning application to processengineering designrdquo Engineering Applications of ArtificialIntelligence vol 41 pp 1ndash16 2015

[17] B Kamsu-Foguem and F H Abanda ldquoExperience modelingwith graphs encoded knowledge for construction industryrdquoComputers in Industry vol 70 pp 79ndash88 2015

[18] B Kamsu-Foguem F H Abanda M B Doumbouya andJ F Tchouanguem ldquoGraph-based ontology reasoning forformal verification of BREEAM rulesrdquo Cognitive SystemsResearch vol 55 pp 14ndash33 2019

[19] S H Othman and G Beydoun ldquoA metamodel-basedknowledge sharing system for disaster managementrdquo ExpertSystems with Applications vol 63 pp 49ndash65 2016

[20] D Xia X Lu H Li W Wang Y Li and Z Zhang ldquoAmapreduce-based parallel frequent pattern growth algorithmfor spatiotemporal association analysis of mobile trajectorybig datardquo Complexity vol 2018 Article ID 2818251 16 pages2018

[21] J Geng S Wang W Gan et al ldquoPromoting geospatial servicefrom information to knowledge with spatiotemporal se-manticsrdquo Complexity vol 2019 Article ID 9301420 14 pages2019

[22] A Nowak-Brzezinska ldquoEnhancing the efficiency of a decisionsupport system through the clustering of complex rule-basedknowledge bases and modification of the inference algo-rithmrdquo Complexity vol 2018 Article ID 2065491 14 pages2018

[23] E Maleki F Belkadi N Boli et al ldquoOntology-basedframework enabling smart product-service systems appli-cation of sensing systems for machine health monitoringrdquoIEEE Internet of ings Journal vol 5 no 6 pp 4496ndash45052018

[24] F H Abanda B Kamsu-Foguem and J H M TahldquoBIMmdashnew rules of measurement ontology for constructioncost estimationrdquo Engineering Science and Technology anInternational Journal vol 20 no 2 pp 443ndash459 2017

[25] B Kamsu-Foguem and P Tiako ldquoRisk information formal-isation with graphsrdquo Computers in Industry vol 85 pp 58ndash69 2017

[26] M Barzegar A Sadeghi-Niaraki and M Shakeri ldquoSpatialexperience based route finding using ontologiesrdquo ETRIJournal pp 1ndash11 2019

[27] E Camossi P Villa and L Mazzola ldquoSemantic-basedanomalous pattern discovery in moving object trajectoriesrdquo2013 httpsarxivorgabs13051946

[28] R Wannous J Malki A Bouju and C Vincent ldquoModellingmobile object activities based on trajectory ontology rulesconsidering spatial relationship rulesrdquo in Modeling Ap-proaches and Algorithms for Advanced Computer Applicationspp 249ndash258 Springer International Publishing BerlinGermany 2013

[29] Y Hu K Janowicz D Carral et al ldquoA geo-ontology designpattern for semantic trajectoriesrdquo in Proceedings of the In-ternational Conference on Spatial Information eorypp 438ndash456 Springer International Publishing ScarboroughUK September 2013

[30] M Baglioni J Macedo C Renso and M Wachowicz ldquoAnontology-based approach for the semantic modelling andreasoning on trajectoriesrdquo in Advances in Conceptual

14 Complexity

converted to an OWL file using OWL Application Pro-gramming Interface (API) in Java Using these routes adriversrsquo experience ontology is separately constructed inProtege software A sample of a route stored in the app isshown in Figure 9

After data preparation the presented algorithm wasimplemented In each run of the app if a new route has beengenerated it will be stored as a class in the driversrsquo expe-rience ontology to be used in future requests without needfor additional processing As a result the routes are stored ina low-volume manner while data processing reduces witheven more runs of the algorithm Each route class in theontology occupies about 20ndash25 percent of its correspondingshapefile Moreover in the presented algorithm two roadnetworks existmdasha peak-time road network and a low-traffic

road network During peak times finding an optimum routebecomes important therefore the algorithm will utilizeDijkstrarsquos algorithm and present the optimum path to theuser

5 Evaluation

In order to evaluate the proposed method 10 different lo-cation-destination pairs covering most of Tehran city wereselected First the experimental road network was fullyconstructed (ER) -en it was constructed only consideringthe low-traffic experimental routes (ETR) Also the shortestpaths between the location-destinations were obtained usingDijkstrarsquos algorithm (DR) For all obtained routes the routelengths were calculated Using driversrsquo reports the travel

Collecting drivers routeswith app

Adding the name (location_destination) of

each new route as a subclass of nontraffic class of experimental

routes

Consider the segments of each path as individuals of

that classrsquo path

Considering the fields (X coordinates Y coordinates)

as the data property

Driversrsquo experienceontology

Experimental route classNonexperimental route class

Adding the name (location_destination) of

each new route as a subclass of nontraffic

class ofnonexperimental routes

Creating experimentaland nonexperimental

route classes

Storing each path in a separate OWL file as a class

Retrieving each class ifrequested by the user

Traffic class of experimental routes

Nontraffic class of experimental routes

Nontraffic class ofnonexperimental routes

Traffic class ofnonexperimental routes

Adding the name (location_destination_t)of each new route as a subclass of traffic class of experimental routes

Is the enteredtime by user is in range

of pick traffic times

Adding the name (location_destination_t)of each new route as a subclass of traffic class

of nonexperimental routes

No Yes

Figure 5 Flowchart of creating driversrsquo experience ontology

8 Complexity

times were also calculated Figure 10 illustrates the resultingroutes for a location-destination pair (Route 10) using thethree methods above

51 Evaluation Based on the Route Length Figure 11 shows adiagram of the length of the paths for the three methods-eroute lengths have been computed by accumulating their

Get the locationdestination and time

from user

Is anyroute with this locationand destination in the

nontraffic classes of driverrsquosexperience ontology

Showing the saved pathto the user

End

Are locationand destinationin experimentalroadnetwork

Using only experimentalroad network for route

finding

Using Dijkstrarsquosalgorithm for route

finding

Finding the nearest point of theexperimental road network to the location

and extraction of a rectangle from themain road network with diameter largerthan the distance between the location

and the nearest point in the experimentalroad network

Is location in experimental

road network

Is destination inexperimental

road network

Route finding with experimentalroad network and segments inthe rectangle(s) of main road

network

Storing route name as a subclass oftraffic class of nonexperimental

road network in driverrsquosexperiences ontology

Finding the nearest point of theexperimental road network to the

destination and extraction of a rectanglefrom the main road network with

diameter larger than the distance betweenthe destination and the nearest point in

the experimental road network

Storing route in aseparate OWL file with

its segments andproperties

Yes

Yes

No

NoNoYes

Is the entered

time in peaktraffic times

Is anyroute with this locationand destination in the

traffic classes of driverrsquosexperience ontology

Considering experimental roadnetwork according to trafficclass of experimental routes

Considering experimental roadnetwork according to nontraffic

class of experimental routes

No Yes

NoNo

Is entered time inpeak traffic times

Storing route name as a subclass ofnontraffic class of nonexperimental

road network indriverrsquos experiences ontology

Yes No

Yes

No

Figure 6 Proposed algorithm flowchart

Complexity 9

constituting segments Figure 12 depicts the ratio of routelengths obtained from the two proposed methods to that ofDijkstrarsquos algorithm Figure 13 compares the mean length ofthe 10 routes for each method

It is observable from the figures that route lengths in theshortest path algorithm are shorter than that of the other twomethods -ey have in contrast greater values when con-sidering high-traffic routes -e mean length of routes is1102 km for Dijkstrarsquos algorithm 1422 km when only con-sidering traffic routes and 134 km when considering all theexperimental (traffic and nontraffic) routes -is is due to taxidrivers who generally select the longest but simultaneously

fastest routes On the other hand Dijkstrarsquos algorithm doesnot consider travel time

52 Evaluation Based on Travel Time Figure 14 shows adiagram of the travel times for the three methods -etravel times have been obtained from different driverswho have driven these routes in peak times and in realconditions Figure 15 depicts the travel times obtainedfrom the two proposed methods to that of Dijkstrarsquos al-gorithm Figure 16 compares the mean travel times foreach method

Experimental road network layer

Basic road network layer

Drsquo

Drsquo

D

Srsquo

SrsquoS

(a)

S

R1

R2

h

D

(b)

Figure 7 (a) Hierarchical experimental road network (b) Extraction of rectangular region from main road network [39]

Trace preprocessing

Ontology design

Ontology-based route finding

Ubiquitous ontology-based route finding

User interface Storingroute if anew routeis created

OWL route filesUser Ontologicalroute-finding

algorithm

GIS and ontologyexperts

Driversrsquo experiences ontology OWL route files Ontology data

Taxiroutedata

Mobile appfor data

collection

Experiencedtaxi drivers

Examiningfrequency ofusing a road

segment

Costmodel

a7

149

10

2 116

15

b

c d

ef

Routenetwork

graph

Figure 8 Steps for implementing the proposed algorithm

10 Complexity

Figure 9 A sample of a stored route in the app which collects data

DR

ETR

ERLocationdestination

Final map

N

S

W E

Figure 10 Resulting routes for a location-destination pair (Route 10) using the three methods

DijkstraExperimental routesExperimental traffic routes

2 3 4 5 6 7 8 9 101Route number

02468

101214161820

Rout

e len

gth

(km

)

Figure 11 Comparison of route length for 10 evaluated routes of the three methods

Complexity 11

It is observable from the figures that travel timesobtained from the hierarchical experimental methodwhen only considering high-traffic conditions are lowesthowever the travel time of the shortest path is the highest-e mean travel time is 80 minutes for Dijkstrarsquos algo-rithm 61 minutes when only considering the high-trafficroutes and 775 minutes when considering all of theexperimental routes Although the routes are long whenconsidering high-traffic routes they are fast due to theselection of low-traffic routes in this case For examplewhen going to Tajrish Square from Valiasr SquareDijkstrarsquos algorithm suggests driving through ValiasrStreet which corresponds to the shortest but not fastestroute However the proposed ontology-based approachsuggests that the driver proceeds mainly along ModarresHighway which reduces the travel time by 35 minutes inhigh-traffic conditions

Another approach to evaluate the two proposedmethodsagainst Dijkstrarsquos algorithm is to find the ideal route amongthem In order to find the ideal route one can divide theroute length by its travel time Obviously when the routelengths for all routes or just two of them are equal the onewhich has the shortest travel time should be selected as theoptimum path On the other hand when the travel times areequal the one which has the shortest length should be se-lected as the optimum path Accordingly in eight cases (of10 cases) the ontology-based route finding considering onlyhigh-traffic routes (ETR) predominates In the other twocases the travel time was roughly equal however the routelengths obtained fromDijkstrarsquos algorithm had lower valuesTo determine the deviation of each method from the idealcase which is shown in Figure 17 the following formula isused

Fj travel length of method j

travel time of method j

j ETR ER Dikjestra

Fe ideal route fraction

Dj Fj minus Fe1113872 1113873

Fe

(3)

-erefore according to the evaluations above it can beclaimed that eliminating the experimental high-traffic routesin high-traffic conditions and utilizing the driversrsquo

02468

10121416

Mea

n of

rout

e len

gth

(km

)

Dijkstra Experimentalroutes

Method name

Experimentaltrafficroutes

Figure 13 Mean of route length for 3 methods

0

20

40

60

80

100

120

Trav

el ti

me (

min

)

1 2 3 4 5 6 7 8 9 10Route number

DijkstraExperimental routesExperimental traffic routes

Figure 14 Comparison of travel time for 10 evaluated routes of 3methods

002040608

112

1 2 3 4 5 6 7 8 9 10

ERDijkstraETRDijkstra

Figure 15 Comparison of travel time for the 2 modes of theproposed method with Dijkstrarsquos algorithm

Dijkstra020406080

Mea

n of

trav

el ti

me (

min

)

Experimentalroutes

Method name

Experimentaltrafficroutes

Figure 16 Mean of travel time for 3 methods

0

05

1

15

2

1 2 3 4 5 6 7 8 9 10

ERDijkstraETRDijkstra

Figure 12 Comparison of route length in 2 modes of the proposedalgorithm with Dijkstrarsquos algorithm

12 Complexity

experiences make the algorithm generate the optimum routethrough low-traffic parts of the network -is minimizestravel time which is the most important parameter in routefinding

6 Conclusions

Different people gain different spatial experiences Asthere are no systems to store these experiences theyevaporate over time-e present study provides the abilityto store spatial experiences and to make the existingsystems in ubiquitous GIS space intelligent In this paperan ontology-based route-finding algorithm in ubiquitousGIS space was designed and implemented using theroutes driven by the drivers of Tehran in high- and low-traffic conditions

In the proposed route-finding algorithm based on thedriversrsquo experiences ontology the number of times that eachsegment is passed by drivers is used to assign weights to thatsegment of the network First the user specifies the location thedestination and the time -en the algorithm determineswhether a subclass with the specified location and destinationexists in the ontology or not If the time chosen falls duringpeak times the existing paths in the traffic subclass are searchedthrough Otherwise the paths in the nontraffic subclass areexamined When a path exists the stored route is displayed tothe user and the algorithm terminates without needing toconduct route finding However when there is no corre-sponding path the algorithm determines whether the locationand destination exist in the experimental road network or notIf both exist in the network only the experimental is used toconstruct the graph and find the route (again peak time

determines the subclass to choosemdashtraffic or nontraffic) andthe algorithm terminates If the origin andor the destinationdoes not exist in the experimental road network the closestpoints of the network to them are found-en a rectangle witha diameter (R2R1 + 2radic2h) greater than the distance betweenthe location (or destination) and the point is extracted from themain road network -e main difference of this algorithm tothe previous experimental route-finding algorithms is the useof ontology and the construction of road networks onlyconsidering the low-traffic paths when there are high-trafficpaths in the network Since each newpath is stored as a separateOWL class the storage issues in the databases are resolved Inaddition the use of ontology causes the system to update everytime the algorithm runs and generates new paths Utilizing onlylow-traffic routes in high-traffic conditions allows the system tosuggest routes that are faster ensuring the optimum path basedon travel time is achieved According to our evaluations theproposed algorithm suggests routes that are longer but si-multaneously the fastest -is guarantees the appropriatenessof using such an algorithm in route finding In this researchany user can access the ontology from any location and at anytime to conduct route finding Moreover being intelligent theubiquitous environment can receive and store the experiences(any data element of ubiquitous GIS) of any person to utilizewithin its services in the future

More innovative applications of spatial experiencemodeling such as ambulance firefighting and tourismservices can be subjects for further considerations -eprocess of route finding for this category is also affected bytraffic as emergency services can also become stuck in trafficlike other drivers because there are no dedicated lines on allroads Due to the limitation in length of road and a largenumber of vehicles in the road in traffic times it is hard forother vehicles to provide a route for traversing of the am-bulance and they still need to find a way to avoid traffic-erefore the experiences from expert technicians in am-bulances or firefighters who have been in real event con-ditions can be modeled and used in the future In additionthis study considers only two factors distance and time forevaluation Other criteria including fuel consumption can beconsidered in future

Data Availability

-e authors are ready to share the research data on request

Disclosure

Maryam Barzegar and Abolghasem Sadeghi-Niaraki are tobe considered as co-first authors

Conflicts of Interest

-e authors declare that they have no conflicts of interest

Acknowledgments

-is research was supported by theMSIT (Ministry of Scienceand ICT) Korea under the ITRC (Information Technology

0

01

02

03

041

2

3

4

5

6

7

8

9

10

DijkstraExperimental routes

Experimental traffic routesIdeal

Figure 17 Comparison of routes from the 3methods with the idealroute

Complexity 13

Research Center) support program (IITP-2019-2016-0-00312)supervised by the IITP (Institute for Information andCommunications Technology Planning and Evaluation)

References

[1] I P Tussyadiah and F J Zach ldquo-e role of geo-basedtechnology in place experiencesrdquo Annals of Tourism Researchvol 39 no 2 pp 780ndash800 2012

[2] G M Honti and J Abonyi ldquoA review of semantic sensortechnologies in internet of things architecturesrdquo Complexityvol 2019 Article ID 6473160 21 pages 2019

[3] U H Govindarajan A J Trappey and C V TrappeyldquoImmersive technology for human-centric cyberphysicalsystems in complex manufacturing processes a compre-hensive overview of the global patent profile using collectiveintelligencerdquo Complexity vol 2018 Article ID 428363417 pages 2018

[4] B K Foguem T Coudert C Beler and L GenesteldquoKnowledge formalization in experience feedback processesan ontology-based approachrdquo Computers in Industry vol 59no 7 pp 694ndash710 2008

[5] N Lasierra F Roldan A Alesanco and J Garcıa ldquoTowardsimproving usage and management of supplies in healthcarean ontology-based solution for sharing knowledgerdquo ExpertSystems with Applications vol 41 no 14 pp 6261ndash6273 2014

[6] M-H Abel ldquoKnowledge map-based web platform to facilitateorganizational learning return of experiencesrdquo Computers inHuman Behavior vol 51 pp 960ndash966 2015

[7] A Garcıa-Crespo J L Lopez-Cuadrado R Colomo-PalaciosI Gonzalez-Carrasco and B Ruiz-Mezcua ldquoSem-fit a se-mantic based expert system to provide recommendations inthe tourism domainrdquo Expert Systems with Applicationsvol 38 no 10 pp 13310ndash13319 2011

[8] D Mourtzis M Doukas and C Giannoulis ldquoAn inference-based knowledge reuse framework for historical product andproduction information retrievalrdquo Procedia CIRP vol 41pp 472ndash477 2016

[9] K Efthymiou K Sipsas D Mourtzis and G ChryssolourisldquoOn knowledge reuse for manufacturing systems design andplanning a semantic technology approachrdquo CIRP Journal ofManufacturing Science and Technology vol 8 pp 1ndash11 2015

[10] W L Mikos J C E Ferreira P E A Botura and L S FreitasldquoA system for distributed sharing and reuse of design andmanufacturing knowledge in the PFMEA domain using adescription logics-based ontologyrdquo Journal of ManufacturingSystems vol 30 no 3 pp 133ndash143 2011

[11] P P Ruiz B K Foguem and B Grabot ldquoGeneratingknowledge in maintenance from experience feedbackrdquoKnowledge-Based Systems vol 68 pp 4ndash20 2014

[12] S Moehrle and W Raskob ldquoStructuring and reusingknowledge from historical events for supporting nuclearemergency and remediation managementrdquo Engineering Ap-plications of Artificial Intelligence vol 46 pp 303ndash311 2015

[13] Q Meng Z Zhang X Wan and X Rong ldquoProperties ex-ploring and information mining in consumer communitynetwork a case of huawei pollen clubrdquo Complexity vol 2018Article ID 9470580 19 pages 2018

[14] R S Renu and GMocko ldquoComputing similarity of text-basedassembly processes for knowledge retrieval and reuserdquoJournal of Manufacturing Systems vol 39 pp 101ndash110 2016

[15] G E Modoni M Doukas W Terkaj M Sacco andD Mourtzis ldquoEnhancing factory data integration through thedevelopment of an ontology from the reference models reuse

to the semantic conversion of the legacy modelsrdquo In-ternational Journal of Computer Integrated Manufacturingvol 30 no 10 pp 1043ndash1059 2017

[16] E R Reyes S Negny G C Robles and J M Le LannldquoImprovement of online adaptation knowledge acquisitionand reuse in case-based reasoning application to processengineering designrdquo Engineering Applications of ArtificialIntelligence vol 41 pp 1ndash16 2015

[17] B Kamsu-Foguem and F H Abanda ldquoExperience modelingwith graphs encoded knowledge for construction industryrdquoComputers in Industry vol 70 pp 79ndash88 2015

[18] B Kamsu-Foguem F H Abanda M B Doumbouya andJ F Tchouanguem ldquoGraph-based ontology reasoning forformal verification of BREEAM rulesrdquo Cognitive SystemsResearch vol 55 pp 14ndash33 2019

[19] S H Othman and G Beydoun ldquoA metamodel-basedknowledge sharing system for disaster managementrdquo ExpertSystems with Applications vol 63 pp 49ndash65 2016

[20] D Xia X Lu H Li W Wang Y Li and Z Zhang ldquoAmapreduce-based parallel frequent pattern growth algorithmfor spatiotemporal association analysis of mobile trajectorybig datardquo Complexity vol 2018 Article ID 2818251 16 pages2018

[21] J Geng S Wang W Gan et al ldquoPromoting geospatial servicefrom information to knowledge with spatiotemporal se-manticsrdquo Complexity vol 2019 Article ID 9301420 14 pages2019

[22] A Nowak-Brzezinska ldquoEnhancing the efficiency of a decisionsupport system through the clustering of complex rule-basedknowledge bases and modification of the inference algo-rithmrdquo Complexity vol 2018 Article ID 2065491 14 pages2018

[23] E Maleki F Belkadi N Boli et al ldquoOntology-basedframework enabling smart product-service systems appli-cation of sensing systems for machine health monitoringrdquoIEEE Internet of ings Journal vol 5 no 6 pp 4496ndash45052018

[24] F H Abanda B Kamsu-Foguem and J H M TahldquoBIMmdashnew rules of measurement ontology for constructioncost estimationrdquo Engineering Science and Technology anInternational Journal vol 20 no 2 pp 443ndash459 2017

[25] B Kamsu-Foguem and P Tiako ldquoRisk information formal-isation with graphsrdquo Computers in Industry vol 85 pp 58ndash69 2017

[26] M Barzegar A Sadeghi-Niaraki and M Shakeri ldquoSpatialexperience based route finding using ontologiesrdquo ETRIJournal pp 1ndash11 2019

[27] E Camossi P Villa and L Mazzola ldquoSemantic-basedanomalous pattern discovery in moving object trajectoriesrdquo2013 httpsarxivorgabs13051946

[28] R Wannous J Malki A Bouju and C Vincent ldquoModellingmobile object activities based on trajectory ontology rulesconsidering spatial relationship rulesrdquo in Modeling Ap-proaches and Algorithms for Advanced Computer Applicationspp 249ndash258 Springer International Publishing BerlinGermany 2013

[29] Y Hu K Janowicz D Carral et al ldquoA geo-ontology designpattern for semantic trajectoriesrdquo in Proceedings of the In-ternational Conference on Spatial Information eorypp 438ndash456 Springer International Publishing ScarboroughUK September 2013

[30] M Baglioni J Macedo C Renso and M Wachowicz ldquoAnontology-based approach for the semantic modelling andreasoning on trajectoriesrdquo in Advances in Conceptual

14 Complexity

times were also calculated Figure 10 illustrates the resultingroutes for a location-destination pair (Route 10) using thethree methods above

51 Evaluation Based on the Route Length Figure 11 shows adiagram of the length of the paths for the three methods-eroute lengths have been computed by accumulating their

Get the locationdestination and time

from user

Is anyroute with this locationand destination in the

nontraffic classes of driverrsquosexperience ontology

Showing the saved pathto the user

End

Are locationand destinationin experimentalroadnetwork

Using only experimentalroad network for route

finding

Using Dijkstrarsquosalgorithm for route

finding

Finding the nearest point of theexperimental road network to the location

and extraction of a rectangle from themain road network with diameter largerthan the distance between the location

and the nearest point in the experimentalroad network

Is location in experimental

road network

Is destination inexperimental

road network

Route finding with experimentalroad network and segments inthe rectangle(s) of main road

network

Storing route name as a subclass oftraffic class of nonexperimental

road network in driverrsquosexperiences ontology

Finding the nearest point of theexperimental road network to the

destination and extraction of a rectanglefrom the main road network with

diameter larger than the distance betweenthe destination and the nearest point in

the experimental road network

Storing route in aseparate OWL file with

its segments andproperties

Yes

Yes

No

NoNoYes

Is the entered

time in peaktraffic times

Is anyroute with this locationand destination in the

traffic classes of driverrsquosexperience ontology

Considering experimental roadnetwork according to trafficclass of experimental routes

Considering experimental roadnetwork according to nontraffic

class of experimental routes

No Yes

NoNo

Is entered time inpeak traffic times

Storing route name as a subclass ofnontraffic class of nonexperimental

road network indriverrsquos experiences ontology

Yes No

Yes

No

Figure 6 Proposed algorithm flowchart

Complexity 9

constituting segments Figure 12 depicts the ratio of routelengths obtained from the two proposed methods to that ofDijkstrarsquos algorithm Figure 13 compares the mean length ofthe 10 routes for each method

It is observable from the figures that route lengths in theshortest path algorithm are shorter than that of the other twomethods -ey have in contrast greater values when con-sidering high-traffic routes -e mean length of routes is1102 km for Dijkstrarsquos algorithm 1422 km when only con-sidering traffic routes and 134 km when considering all theexperimental (traffic and nontraffic) routes -is is due to taxidrivers who generally select the longest but simultaneously

fastest routes On the other hand Dijkstrarsquos algorithm doesnot consider travel time

52 Evaluation Based on Travel Time Figure 14 shows adiagram of the travel times for the three methods -etravel times have been obtained from different driverswho have driven these routes in peak times and in realconditions Figure 15 depicts the travel times obtainedfrom the two proposed methods to that of Dijkstrarsquos al-gorithm Figure 16 compares the mean travel times foreach method

Experimental road network layer

Basic road network layer

Drsquo

Drsquo

D

Srsquo

SrsquoS

(a)

S

R1

R2

h

D

(b)

Figure 7 (a) Hierarchical experimental road network (b) Extraction of rectangular region from main road network [39]

Trace preprocessing

Ontology design

Ontology-based route finding

Ubiquitous ontology-based route finding

User interface Storingroute if anew routeis created

OWL route filesUser Ontologicalroute-finding

algorithm

GIS and ontologyexperts

Driversrsquo experiences ontology OWL route files Ontology data

Taxiroutedata

Mobile appfor data

collection

Experiencedtaxi drivers

Examiningfrequency ofusing a road

segment

Costmodel

a7

149

10

2 116

15

b

c d

ef

Routenetwork

graph

Figure 8 Steps for implementing the proposed algorithm

10 Complexity

Figure 9 A sample of a stored route in the app which collects data

DR

ETR

ERLocationdestination

Final map

N

S

W E

Figure 10 Resulting routes for a location-destination pair (Route 10) using the three methods

DijkstraExperimental routesExperimental traffic routes

2 3 4 5 6 7 8 9 101Route number

02468

101214161820

Rout

e len

gth

(km

)

Figure 11 Comparison of route length for 10 evaluated routes of the three methods

Complexity 11

It is observable from the figures that travel timesobtained from the hierarchical experimental methodwhen only considering high-traffic conditions are lowesthowever the travel time of the shortest path is the highest-e mean travel time is 80 minutes for Dijkstrarsquos algo-rithm 61 minutes when only considering the high-trafficroutes and 775 minutes when considering all of theexperimental routes Although the routes are long whenconsidering high-traffic routes they are fast due to theselection of low-traffic routes in this case For examplewhen going to Tajrish Square from Valiasr SquareDijkstrarsquos algorithm suggests driving through ValiasrStreet which corresponds to the shortest but not fastestroute However the proposed ontology-based approachsuggests that the driver proceeds mainly along ModarresHighway which reduces the travel time by 35 minutes inhigh-traffic conditions

Another approach to evaluate the two proposedmethodsagainst Dijkstrarsquos algorithm is to find the ideal route amongthem In order to find the ideal route one can divide theroute length by its travel time Obviously when the routelengths for all routes or just two of them are equal the onewhich has the shortest travel time should be selected as theoptimum path On the other hand when the travel times areequal the one which has the shortest length should be se-lected as the optimum path Accordingly in eight cases (of10 cases) the ontology-based route finding considering onlyhigh-traffic routes (ETR) predominates In the other twocases the travel time was roughly equal however the routelengths obtained fromDijkstrarsquos algorithm had lower valuesTo determine the deviation of each method from the idealcase which is shown in Figure 17 the following formula isused

Fj travel length of method j

travel time of method j

j ETR ER Dikjestra

Fe ideal route fraction

Dj Fj minus Fe1113872 1113873

Fe

(3)

-erefore according to the evaluations above it can beclaimed that eliminating the experimental high-traffic routesin high-traffic conditions and utilizing the driversrsquo

02468

10121416

Mea

n of

rout

e len

gth

(km

)

Dijkstra Experimentalroutes

Method name

Experimentaltrafficroutes

Figure 13 Mean of route length for 3 methods

0

20

40

60

80

100

120

Trav

el ti

me (

min

)

1 2 3 4 5 6 7 8 9 10Route number

DijkstraExperimental routesExperimental traffic routes

Figure 14 Comparison of travel time for 10 evaluated routes of 3methods

002040608

112

1 2 3 4 5 6 7 8 9 10

ERDijkstraETRDijkstra

Figure 15 Comparison of travel time for the 2 modes of theproposed method with Dijkstrarsquos algorithm

Dijkstra020406080

Mea

n of

trav

el ti

me (

min

)

Experimentalroutes

Method name

Experimentaltrafficroutes

Figure 16 Mean of travel time for 3 methods

0

05

1

15

2

1 2 3 4 5 6 7 8 9 10

ERDijkstraETRDijkstra

Figure 12 Comparison of route length in 2 modes of the proposedalgorithm with Dijkstrarsquos algorithm

12 Complexity

experiences make the algorithm generate the optimum routethrough low-traffic parts of the network -is minimizestravel time which is the most important parameter in routefinding

6 Conclusions

Different people gain different spatial experiences Asthere are no systems to store these experiences theyevaporate over time-e present study provides the abilityto store spatial experiences and to make the existingsystems in ubiquitous GIS space intelligent In this paperan ontology-based route-finding algorithm in ubiquitousGIS space was designed and implemented using theroutes driven by the drivers of Tehran in high- and low-traffic conditions

In the proposed route-finding algorithm based on thedriversrsquo experiences ontology the number of times that eachsegment is passed by drivers is used to assign weights to thatsegment of the network First the user specifies the location thedestination and the time -en the algorithm determineswhether a subclass with the specified location and destinationexists in the ontology or not If the time chosen falls duringpeak times the existing paths in the traffic subclass are searchedthrough Otherwise the paths in the nontraffic subclass areexamined When a path exists the stored route is displayed tothe user and the algorithm terminates without needing toconduct route finding However when there is no corre-sponding path the algorithm determines whether the locationand destination exist in the experimental road network or notIf both exist in the network only the experimental is used toconstruct the graph and find the route (again peak time

determines the subclass to choosemdashtraffic or nontraffic) andthe algorithm terminates If the origin andor the destinationdoes not exist in the experimental road network the closestpoints of the network to them are found-en a rectangle witha diameter (R2R1 + 2radic2h) greater than the distance betweenthe location (or destination) and the point is extracted from themain road network -e main difference of this algorithm tothe previous experimental route-finding algorithms is the useof ontology and the construction of road networks onlyconsidering the low-traffic paths when there are high-trafficpaths in the network Since each newpath is stored as a separateOWL class the storage issues in the databases are resolved Inaddition the use of ontology causes the system to update everytime the algorithm runs and generates new paths Utilizing onlylow-traffic routes in high-traffic conditions allows the system tosuggest routes that are faster ensuring the optimum path basedon travel time is achieved According to our evaluations theproposed algorithm suggests routes that are longer but si-multaneously the fastest -is guarantees the appropriatenessof using such an algorithm in route finding In this researchany user can access the ontology from any location and at anytime to conduct route finding Moreover being intelligent theubiquitous environment can receive and store the experiences(any data element of ubiquitous GIS) of any person to utilizewithin its services in the future

More innovative applications of spatial experiencemodeling such as ambulance firefighting and tourismservices can be subjects for further considerations -eprocess of route finding for this category is also affected bytraffic as emergency services can also become stuck in trafficlike other drivers because there are no dedicated lines on allroads Due to the limitation in length of road and a largenumber of vehicles in the road in traffic times it is hard forother vehicles to provide a route for traversing of the am-bulance and they still need to find a way to avoid traffic-erefore the experiences from expert technicians in am-bulances or firefighters who have been in real event con-ditions can be modeled and used in the future In additionthis study considers only two factors distance and time forevaluation Other criteria including fuel consumption can beconsidered in future

Data Availability

-e authors are ready to share the research data on request

Disclosure

Maryam Barzegar and Abolghasem Sadeghi-Niaraki are tobe considered as co-first authors

Conflicts of Interest

-e authors declare that they have no conflicts of interest

Acknowledgments

-is research was supported by theMSIT (Ministry of Scienceand ICT) Korea under the ITRC (Information Technology

0

01

02

03

041

2

3

4

5

6

7

8

9

10

DijkstraExperimental routes

Experimental traffic routesIdeal

Figure 17 Comparison of routes from the 3methods with the idealroute

Complexity 13

Research Center) support program (IITP-2019-2016-0-00312)supervised by the IITP (Institute for Information andCommunications Technology Planning and Evaluation)

References

[1] I P Tussyadiah and F J Zach ldquo-e role of geo-basedtechnology in place experiencesrdquo Annals of Tourism Researchvol 39 no 2 pp 780ndash800 2012

[2] G M Honti and J Abonyi ldquoA review of semantic sensortechnologies in internet of things architecturesrdquo Complexityvol 2019 Article ID 6473160 21 pages 2019

[3] U H Govindarajan A J Trappey and C V TrappeyldquoImmersive technology for human-centric cyberphysicalsystems in complex manufacturing processes a compre-hensive overview of the global patent profile using collectiveintelligencerdquo Complexity vol 2018 Article ID 428363417 pages 2018

[4] B K Foguem T Coudert C Beler and L GenesteldquoKnowledge formalization in experience feedback processesan ontology-based approachrdquo Computers in Industry vol 59no 7 pp 694ndash710 2008

[5] N Lasierra F Roldan A Alesanco and J Garcıa ldquoTowardsimproving usage and management of supplies in healthcarean ontology-based solution for sharing knowledgerdquo ExpertSystems with Applications vol 41 no 14 pp 6261ndash6273 2014

[6] M-H Abel ldquoKnowledge map-based web platform to facilitateorganizational learning return of experiencesrdquo Computers inHuman Behavior vol 51 pp 960ndash966 2015

[7] A Garcıa-Crespo J L Lopez-Cuadrado R Colomo-PalaciosI Gonzalez-Carrasco and B Ruiz-Mezcua ldquoSem-fit a se-mantic based expert system to provide recommendations inthe tourism domainrdquo Expert Systems with Applicationsvol 38 no 10 pp 13310ndash13319 2011

[8] D Mourtzis M Doukas and C Giannoulis ldquoAn inference-based knowledge reuse framework for historical product andproduction information retrievalrdquo Procedia CIRP vol 41pp 472ndash477 2016

[9] K Efthymiou K Sipsas D Mourtzis and G ChryssolourisldquoOn knowledge reuse for manufacturing systems design andplanning a semantic technology approachrdquo CIRP Journal ofManufacturing Science and Technology vol 8 pp 1ndash11 2015

[10] W L Mikos J C E Ferreira P E A Botura and L S FreitasldquoA system for distributed sharing and reuse of design andmanufacturing knowledge in the PFMEA domain using adescription logics-based ontologyrdquo Journal of ManufacturingSystems vol 30 no 3 pp 133ndash143 2011

[11] P P Ruiz B K Foguem and B Grabot ldquoGeneratingknowledge in maintenance from experience feedbackrdquoKnowledge-Based Systems vol 68 pp 4ndash20 2014

[12] S Moehrle and W Raskob ldquoStructuring and reusingknowledge from historical events for supporting nuclearemergency and remediation managementrdquo Engineering Ap-plications of Artificial Intelligence vol 46 pp 303ndash311 2015

[13] Q Meng Z Zhang X Wan and X Rong ldquoProperties ex-ploring and information mining in consumer communitynetwork a case of huawei pollen clubrdquo Complexity vol 2018Article ID 9470580 19 pages 2018

[14] R S Renu and GMocko ldquoComputing similarity of text-basedassembly processes for knowledge retrieval and reuserdquoJournal of Manufacturing Systems vol 39 pp 101ndash110 2016

[15] G E Modoni M Doukas W Terkaj M Sacco andD Mourtzis ldquoEnhancing factory data integration through thedevelopment of an ontology from the reference models reuse

to the semantic conversion of the legacy modelsrdquo In-ternational Journal of Computer Integrated Manufacturingvol 30 no 10 pp 1043ndash1059 2017

[16] E R Reyes S Negny G C Robles and J M Le LannldquoImprovement of online adaptation knowledge acquisitionand reuse in case-based reasoning application to processengineering designrdquo Engineering Applications of ArtificialIntelligence vol 41 pp 1ndash16 2015

[17] B Kamsu-Foguem and F H Abanda ldquoExperience modelingwith graphs encoded knowledge for construction industryrdquoComputers in Industry vol 70 pp 79ndash88 2015

[18] B Kamsu-Foguem F H Abanda M B Doumbouya andJ F Tchouanguem ldquoGraph-based ontology reasoning forformal verification of BREEAM rulesrdquo Cognitive SystemsResearch vol 55 pp 14ndash33 2019

[19] S H Othman and G Beydoun ldquoA metamodel-basedknowledge sharing system for disaster managementrdquo ExpertSystems with Applications vol 63 pp 49ndash65 2016

[20] D Xia X Lu H Li W Wang Y Li and Z Zhang ldquoAmapreduce-based parallel frequent pattern growth algorithmfor spatiotemporal association analysis of mobile trajectorybig datardquo Complexity vol 2018 Article ID 2818251 16 pages2018

[21] J Geng S Wang W Gan et al ldquoPromoting geospatial servicefrom information to knowledge with spatiotemporal se-manticsrdquo Complexity vol 2019 Article ID 9301420 14 pages2019

[22] A Nowak-Brzezinska ldquoEnhancing the efficiency of a decisionsupport system through the clustering of complex rule-basedknowledge bases and modification of the inference algo-rithmrdquo Complexity vol 2018 Article ID 2065491 14 pages2018

[23] E Maleki F Belkadi N Boli et al ldquoOntology-basedframework enabling smart product-service systems appli-cation of sensing systems for machine health monitoringrdquoIEEE Internet of ings Journal vol 5 no 6 pp 4496ndash45052018

[24] F H Abanda B Kamsu-Foguem and J H M TahldquoBIMmdashnew rules of measurement ontology for constructioncost estimationrdquo Engineering Science and Technology anInternational Journal vol 20 no 2 pp 443ndash459 2017

[25] B Kamsu-Foguem and P Tiako ldquoRisk information formal-isation with graphsrdquo Computers in Industry vol 85 pp 58ndash69 2017

[26] M Barzegar A Sadeghi-Niaraki and M Shakeri ldquoSpatialexperience based route finding using ontologiesrdquo ETRIJournal pp 1ndash11 2019

[27] E Camossi P Villa and L Mazzola ldquoSemantic-basedanomalous pattern discovery in moving object trajectoriesrdquo2013 httpsarxivorgabs13051946

[28] R Wannous J Malki A Bouju and C Vincent ldquoModellingmobile object activities based on trajectory ontology rulesconsidering spatial relationship rulesrdquo in Modeling Ap-proaches and Algorithms for Advanced Computer Applicationspp 249ndash258 Springer International Publishing BerlinGermany 2013

[29] Y Hu K Janowicz D Carral et al ldquoA geo-ontology designpattern for semantic trajectoriesrdquo in Proceedings of the In-ternational Conference on Spatial Information eorypp 438ndash456 Springer International Publishing ScarboroughUK September 2013

[30] M Baglioni J Macedo C Renso and M Wachowicz ldquoAnontology-based approach for the semantic modelling andreasoning on trajectoriesrdquo in Advances in Conceptual

14 Complexity

constituting segments Figure 12 depicts the ratio of routelengths obtained from the two proposed methods to that ofDijkstrarsquos algorithm Figure 13 compares the mean length ofthe 10 routes for each method

It is observable from the figures that route lengths in theshortest path algorithm are shorter than that of the other twomethods -ey have in contrast greater values when con-sidering high-traffic routes -e mean length of routes is1102 km for Dijkstrarsquos algorithm 1422 km when only con-sidering traffic routes and 134 km when considering all theexperimental (traffic and nontraffic) routes -is is due to taxidrivers who generally select the longest but simultaneously

fastest routes On the other hand Dijkstrarsquos algorithm doesnot consider travel time

52 Evaluation Based on Travel Time Figure 14 shows adiagram of the travel times for the three methods -etravel times have been obtained from different driverswho have driven these routes in peak times and in realconditions Figure 15 depicts the travel times obtainedfrom the two proposed methods to that of Dijkstrarsquos al-gorithm Figure 16 compares the mean travel times foreach method

Experimental road network layer

Basic road network layer

Drsquo

Drsquo

D

Srsquo

SrsquoS

(a)

S

R1

R2

h

D

(b)

Figure 7 (a) Hierarchical experimental road network (b) Extraction of rectangular region from main road network [39]

Trace preprocessing

Ontology design

Ontology-based route finding

Ubiquitous ontology-based route finding

User interface Storingroute if anew routeis created

OWL route filesUser Ontologicalroute-finding

algorithm

GIS and ontologyexperts

Driversrsquo experiences ontology OWL route files Ontology data

Taxiroutedata

Mobile appfor data

collection

Experiencedtaxi drivers

Examiningfrequency ofusing a road

segment

Costmodel

a7

149

10

2 116

15

b

c d

ef

Routenetwork

graph

Figure 8 Steps for implementing the proposed algorithm

10 Complexity

Figure 9 A sample of a stored route in the app which collects data

DR

ETR

ERLocationdestination

Final map

N

S

W E

Figure 10 Resulting routes for a location-destination pair (Route 10) using the three methods

DijkstraExperimental routesExperimental traffic routes

2 3 4 5 6 7 8 9 101Route number

02468

101214161820

Rout

e len

gth

(km

)

Figure 11 Comparison of route length for 10 evaluated routes of the three methods

Complexity 11

It is observable from the figures that travel timesobtained from the hierarchical experimental methodwhen only considering high-traffic conditions are lowesthowever the travel time of the shortest path is the highest-e mean travel time is 80 minutes for Dijkstrarsquos algo-rithm 61 minutes when only considering the high-trafficroutes and 775 minutes when considering all of theexperimental routes Although the routes are long whenconsidering high-traffic routes they are fast due to theselection of low-traffic routes in this case For examplewhen going to Tajrish Square from Valiasr SquareDijkstrarsquos algorithm suggests driving through ValiasrStreet which corresponds to the shortest but not fastestroute However the proposed ontology-based approachsuggests that the driver proceeds mainly along ModarresHighway which reduces the travel time by 35 minutes inhigh-traffic conditions

Another approach to evaluate the two proposedmethodsagainst Dijkstrarsquos algorithm is to find the ideal route amongthem In order to find the ideal route one can divide theroute length by its travel time Obviously when the routelengths for all routes or just two of them are equal the onewhich has the shortest travel time should be selected as theoptimum path On the other hand when the travel times areequal the one which has the shortest length should be se-lected as the optimum path Accordingly in eight cases (of10 cases) the ontology-based route finding considering onlyhigh-traffic routes (ETR) predominates In the other twocases the travel time was roughly equal however the routelengths obtained fromDijkstrarsquos algorithm had lower valuesTo determine the deviation of each method from the idealcase which is shown in Figure 17 the following formula isused

Fj travel length of method j

travel time of method j

j ETR ER Dikjestra

Fe ideal route fraction

Dj Fj minus Fe1113872 1113873

Fe

(3)

-erefore according to the evaluations above it can beclaimed that eliminating the experimental high-traffic routesin high-traffic conditions and utilizing the driversrsquo

02468

10121416

Mea

n of

rout

e len

gth

(km

)

Dijkstra Experimentalroutes

Method name

Experimentaltrafficroutes

Figure 13 Mean of route length for 3 methods

0

20

40

60

80

100

120

Trav

el ti

me (

min

)

1 2 3 4 5 6 7 8 9 10Route number

DijkstraExperimental routesExperimental traffic routes

Figure 14 Comparison of travel time for 10 evaluated routes of 3methods

002040608

112

1 2 3 4 5 6 7 8 9 10

ERDijkstraETRDijkstra

Figure 15 Comparison of travel time for the 2 modes of theproposed method with Dijkstrarsquos algorithm

Dijkstra020406080

Mea

n of

trav

el ti

me (

min

)

Experimentalroutes

Method name

Experimentaltrafficroutes

Figure 16 Mean of travel time for 3 methods

0

05

1

15

2

1 2 3 4 5 6 7 8 9 10

ERDijkstraETRDijkstra

Figure 12 Comparison of route length in 2 modes of the proposedalgorithm with Dijkstrarsquos algorithm

12 Complexity

experiences make the algorithm generate the optimum routethrough low-traffic parts of the network -is minimizestravel time which is the most important parameter in routefinding

6 Conclusions

Different people gain different spatial experiences Asthere are no systems to store these experiences theyevaporate over time-e present study provides the abilityto store spatial experiences and to make the existingsystems in ubiquitous GIS space intelligent In this paperan ontology-based route-finding algorithm in ubiquitousGIS space was designed and implemented using theroutes driven by the drivers of Tehran in high- and low-traffic conditions

In the proposed route-finding algorithm based on thedriversrsquo experiences ontology the number of times that eachsegment is passed by drivers is used to assign weights to thatsegment of the network First the user specifies the location thedestination and the time -en the algorithm determineswhether a subclass with the specified location and destinationexists in the ontology or not If the time chosen falls duringpeak times the existing paths in the traffic subclass are searchedthrough Otherwise the paths in the nontraffic subclass areexamined When a path exists the stored route is displayed tothe user and the algorithm terminates without needing toconduct route finding However when there is no corre-sponding path the algorithm determines whether the locationand destination exist in the experimental road network or notIf both exist in the network only the experimental is used toconstruct the graph and find the route (again peak time

determines the subclass to choosemdashtraffic or nontraffic) andthe algorithm terminates If the origin andor the destinationdoes not exist in the experimental road network the closestpoints of the network to them are found-en a rectangle witha diameter (R2R1 + 2radic2h) greater than the distance betweenthe location (or destination) and the point is extracted from themain road network -e main difference of this algorithm tothe previous experimental route-finding algorithms is the useof ontology and the construction of road networks onlyconsidering the low-traffic paths when there are high-trafficpaths in the network Since each newpath is stored as a separateOWL class the storage issues in the databases are resolved Inaddition the use of ontology causes the system to update everytime the algorithm runs and generates new paths Utilizing onlylow-traffic routes in high-traffic conditions allows the system tosuggest routes that are faster ensuring the optimum path basedon travel time is achieved According to our evaluations theproposed algorithm suggests routes that are longer but si-multaneously the fastest -is guarantees the appropriatenessof using such an algorithm in route finding In this researchany user can access the ontology from any location and at anytime to conduct route finding Moreover being intelligent theubiquitous environment can receive and store the experiences(any data element of ubiquitous GIS) of any person to utilizewithin its services in the future

More innovative applications of spatial experiencemodeling such as ambulance firefighting and tourismservices can be subjects for further considerations -eprocess of route finding for this category is also affected bytraffic as emergency services can also become stuck in trafficlike other drivers because there are no dedicated lines on allroads Due to the limitation in length of road and a largenumber of vehicles in the road in traffic times it is hard forother vehicles to provide a route for traversing of the am-bulance and they still need to find a way to avoid traffic-erefore the experiences from expert technicians in am-bulances or firefighters who have been in real event con-ditions can be modeled and used in the future In additionthis study considers only two factors distance and time forevaluation Other criteria including fuel consumption can beconsidered in future

Data Availability

-e authors are ready to share the research data on request

Disclosure

Maryam Barzegar and Abolghasem Sadeghi-Niaraki are tobe considered as co-first authors

Conflicts of Interest

-e authors declare that they have no conflicts of interest

Acknowledgments

-is research was supported by theMSIT (Ministry of Scienceand ICT) Korea under the ITRC (Information Technology

0

01

02

03

041

2

3

4

5

6

7

8

9

10

DijkstraExperimental routes

Experimental traffic routesIdeal

Figure 17 Comparison of routes from the 3methods with the idealroute

Complexity 13

Research Center) support program (IITP-2019-2016-0-00312)supervised by the IITP (Institute for Information andCommunications Technology Planning and Evaluation)

References

[1] I P Tussyadiah and F J Zach ldquo-e role of geo-basedtechnology in place experiencesrdquo Annals of Tourism Researchvol 39 no 2 pp 780ndash800 2012

[2] G M Honti and J Abonyi ldquoA review of semantic sensortechnologies in internet of things architecturesrdquo Complexityvol 2019 Article ID 6473160 21 pages 2019

[3] U H Govindarajan A J Trappey and C V TrappeyldquoImmersive technology for human-centric cyberphysicalsystems in complex manufacturing processes a compre-hensive overview of the global patent profile using collectiveintelligencerdquo Complexity vol 2018 Article ID 428363417 pages 2018

[4] B K Foguem T Coudert C Beler and L GenesteldquoKnowledge formalization in experience feedback processesan ontology-based approachrdquo Computers in Industry vol 59no 7 pp 694ndash710 2008

[5] N Lasierra F Roldan A Alesanco and J Garcıa ldquoTowardsimproving usage and management of supplies in healthcarean ontology-based solution for sharing knowledgerdquo ExpertSystems with Applications vol 41 no 14 pp 6261ndash6273 2014

[6] M-H Abel ldquoKnowledge map-based web platform to facilitateorganizational learning return of experiencesrdquo Computers inHuman Behavior vol 51 pp 960ndash966 2015

[7] A Garcıa-Crespo J L Lopez-Cuadrado R Colomo-PalaciosI Gonzalez-Carrasco and B Ruiz-Mezcua ldquoSem-fit a se-mantic based expert system to provide recommendations inthe tourism domainrdquo Expert Systems with Applicationsvol 38 no 10 pp 13310ndash13319 2011

[8] D Mourtzis M Doukas and C Giannoulis ldquoAn inference-based knowledge reuse framework for historical product andproduction information retrievalrdquo Procedia CIRP vol 41pp 472ndash477 2016

[9] K Efthymiou K Sipsas D Mourtzis and G ChryssolourisldquoOn knowledge reuse for manufacturing systems design andplanning a semantic technology approachrdquo CIRP Journal ofManufacturing Science and Technology vol 8 pp 1ndash11 2015

[10] W L Mikos J C E Ferreira P E A Botura and L S FreitasldquoA system for distributed sharing and reuse of design andmanufacturing knowledge in the PFMEA domain using adescription logics-based ontologyrdquo Journal of ManufacturingSystems vol 30 no 3 pp 133ndash143 2011

[11] P P Ruiz B K Foguem and B Grabot ldquoGeneratingknowledge in maintenance from experience feedbackrdquoKnowledge-Based Systems vol 68 pp 4ndash20 2014

[12] S Moehrle and W Raskob ldquoStructuring and reusingknowledge from historical events for supporting nuclearemergency and remediation managementrdquo Engineering Ap-plications of Artificial Intelligence vol 46 pp 303ndash311 2015

[13] Q Meng Z Zhang X Wan and X Rong ldquoProperties ex-ploring and information mining in consumer communitynetwork a case of huawei pollen clubrdquo Complexity vol 2018Article ID 9470580 19 pages 2018

[14] R S Renu and GMocko ldquoComputing similarity of text-basedassembly processes for knowledge retrieval and reuserdquoJournal of Manufacturing Systems vol 39 pp 101ndash110 2016

[15] G E Modoni M Doukas W Terkaj M Sacco andD Mourtzis ldquoEnhancing factory data integration through thedevelopment of an ontology from the reference models reuse

to the semantic conversion of the legacy modelsrdquo In-ternational Journal of Computer Integrated Manufacturingvol 30 no 10 pp 1043ndash1059 2017

[16] E R Reyes S Negny G C Robles and J M Le LannldquoImprovement of online adaptation knowledge acquisitionand reuse in case-based reasoning application to processengineering designrdquo Engineering Applications of ArtificialIntelligence vol 41 pp 1ndash16 2015

[17] B Kamsu-Foguem and F H Abanda ldquoExperience modelingwith graphs encoded knowledge for construction industryrdquoComputers in Industry vol 70 pp 79ndash88 2015

[18] B Kamsu-Foguem F H Abanda M B Doumbouya andJ F Tchouanguem ldquoGraph-based ontology reasoning forformal verification of BREEAM rulesrdquo Cognitive SystemsResearch vol 55 pp 14ndash33 2019

[19] S H Othman and G Beydoun ldquoA metamodel-basedknowledge sharing system for disaster managementrdquo ExpertSystems with Applications vol 63 pp 49ndash65 2016

[20] D Xia X Lu H Li W Wang Y Li and Z Zhang ldquoAmapreduce-based parallel frequent pattern growth algorithmfor spatiotemporal association analysis of mobile trajectorybig datardquo Complexity vol 2018 Article ID 2818251 16 pages2018

[21] J Geng S Wang W Gan et al ldquoPromoting geospatial servicefrom information to knowledge with spatiotemporal se-manticsrdquo Complexity vol 2019 Article ID 9301420 14 pages2019

[22] A Nowak-Brzezinska ldquoEnhancing the efficiency of a decisionsupport system through the clustering of complex rule-basedknowledge bases and modification of the inference algo-rithmrdquo Complexity vol 2018 Article ID 2065491 14 pages2018

[23] E Maleki F Belkadi N Boli et al ldquoOntology-basedframework enabling smart product-service systems appli-cation of sensing systems for machine health monitoringrdquoIEEE Internet of ings Journal vol 5 no 6 pp 4496ndash45052018

[24] F H Abanda B Kamsu-Foguem and J H M TahldquoBIMmdashnew rules of measurement ontology for constructioncost estimationrdquo Engineering Science and Technology anInternational Journal vol 20 no 2 pp 443ndash459 2017

[25] B Kamsu-Foguem and P Tiako ldquoRisk information formal-isation with graphsrdquo Computers in Industry vol 85 pp 58ndash69 2017

[26] M Barzegar A Sadeghi-Niaraki and M Shakeri ldquoSpatialexperience based route finding using ontologiesrdquo ETRIJournal pp 1ndash11 2019

[27] E Camossi P Villa and L Mazzola ldquoSemantic-basedanomalous pattern discovery in moving object trajectoriesrdquo2013 httpsarxivorgabs13051946

[28] R Wannous J Malki A Bouju and C Vincent ldquoModellingmobile object activities based on trajectory ontology rulesconsidering spatial relationship rulesrdquo in Modeling Ap-proaches and Algorithms for Advanced Computer Applicationspp 249ndash258 Springer International Publishing BerlinGermany 2013

[29] Y Hu K Janowicz D Carral et al ldquoA geo-ontology designpattern for semantic trajectoriesrdquo in Proceedings of the In-ternational Conference on Spatial Information eorypp 438ndash456 Springer International Publishing ScarboroughUK September 2013

[30] M Baglioni J Macedo C Renso and M Wachowicz ldquoAnontology-based approach for the semantic modelling andreasoning on trajectoriesrdquo in Advances in Conceptual

14 Complexity

Figure 9 A sample of a stored route in the app which collects data

DR

ETR

ERLocationdestination

Final map

N

S

W E

Figure 10 Resulting routes for a location-destination pair (Route 10) using the three methods

DijkstraExperimental routesExperimental traffic routes

2 3 4 5 6 7 8 9 101Route number

02468

101214161820

Rout

e len

gth

(km

)

Figure 11 Comparison of route length for 10 evaluated routes of the three methods

Complexity 11

It is observable from the figures that travel timesobtained from the hierarchical experimental methodwhen only considering high-traffic conditions are lowesthowever the travel time of the shortest path is the highest-e mean travel time is 80 minutes for Dijkstrarsquos algo-rithm 61 minutes when only considering the high-trafficroutes and 775 minutes when considering all of theexperimental routes Although the routes are long whenconsidering high-traffic routes they are fast due to theselection of low-traffic routes in this case For examplewhen going to Tajrish Square from Valiasr SquareDijkstrarsquos algorithm suggests driving through ValiasrStreet which corresponds to the shortest but not fastestroute However the proposed ontology-based approachsuggests that the driver proceeds mainly along ModarresHighway which reduces the travel time by 35 minutes inhigh-traffic conditions

Another approach to evaluate the two proposedmethodsagainst Dijkstrarsquos algorithm is to find the ideal route amongthem In order to find the ideal route one can divide theroute length by its travel time Obviously when the routelengths for all routes or just two of them are equal the onewhich has the shortest travel time should be selected as theoptimum path On the other hand when the travel times areequal the one which has the shortest length should be se-lected as the optimum path Accordingly in eight cases (of10 cases) the ontology-based route finding considering onlyhigh-traffic routes (ETR) predominates In the other twocases the travel time was roughly equal however the routelengths obtained fromDijkstrarsquos algorithm had lower valuesTo determine the deviation of each method from the idealcase which is shown in Figure 17 the following formula isused

Fj travel length of method j

travel time of method j

j ETR ER Dikjestra

Fe ideal route fraction

Dj Fj minus Fe1113872 1113873

Fe

(3)

-erefore according to the evaluations above it can beclaimed that eliminating the experimental high-traffic routesin high-traffic conditions and utilizing the driversrsquo

02468

10121416

Mea

n of

rout

e len

gth

(km

)

Dijkstra Experimentalroutes

Method name

Experimentaltrafficroutes

Figure 13 Mean of route length for 3 methods

0

20

40

60

80

100

120

Trav

el ti

me (

min

)

1 2 3 4 5 6 7 8 9 10Route number

DijkstraExperimental routesExperimental traffic routes

Figure 14 Comparison of travel time for 10 evaluated routes of 3methods

002040608

112

1 2 3 4 5 6 7 8 9 10

ERDijkstraETRDijkstra

Figure 15 Comparison of travel time for the 2 modes of theproposed method with Dijkstrarsquos algorithm

Dijkstra020406080

Mea

n of

trav

el ti

me (

min

)

Experimentalroutes

Method name

Experimentaltrafficroutes

Figure 16 Mean of travel time for 3 methods

0

05

1

15

2

1 2 3 4 5 6 7 8 9 10

ERDijkstraETRDijkstra

Figure 12 Comparison of route length in 2 modes of the proposedalgorithm with Dijkstrarsquos algorithm

12 Complexity

experiences make the algorithm generate the optimum routethrough low-traffic parts of the network -is minimizestravel time which is the most important parameter in routefinding

6 Conclusions

Different people gain different spatial experiences Asthere are no systems to store these experiences theyevaporate over time-e present study provides the abilityto store spatial experiences and to make the existingsystems in ubiquitous GIS space intelligent In this paperan ontology-based route-finding algorithm in ubiquitousGIS space was designed and implemented using theroutes driven by the drivers of Tehran in high- and low-traffic conditions

In the proposed route-finding algorithm based on thedriversrsquo experiences ontology the number of times that eachsegment is passed by drivers is used to assign weights to thatsegment of the network First the user specifies the location thedestination and the time -en the algorithm determineswhether a subclass with the specified location and destinationexists in the ontology or not If the time chosen falls duringpeak times the existing paths in the traffic subclass are searchedthrough Otherwise the paths in the nontraffic subclass areexamined When a path exists the stored route is displayed tothe user and the algorithm terminates without needing toconduct route finding However when there is no corre-sponding path the algorithm determines whether the locationand destination exist in the experimental road network or notIf both exist in the network only the experimental is used toconstruct the graph and find the route (again peak time

determines the subclass to choosemdashtraffic or nontraffic) andthe algorithm terminates If the origin andor the destinationdoes not exist in the experimental road network the closestpoints of the network to them are found-en a rectangle witha diameter (R2R1 + 2radic2h) greater than the distance betweenthe location (or destination) and the point is extracted from themain road network -e main difference of this algorithm tothe previous experimental route-finding algorithms is the useof ontology and the construction of road networks onlyconsidering the low-traffic paths when there are high-trafficpaths in the network Since each newpath is stored as a separateOWL class the storage issues in the databases are resolved Inaddition the use of ontology causes the system to update everytime the algorithm runs and generates new paths Utilizing onlylow-traffic routes in high-traffic conditions allows the system tosuggest routes that are faster ensuring the optimum path basedon travel time is achieved According to our evaluations theproposed algorithm suggests routes that are longer but si-multaneously the fastest -is guarantees the appropriatenessof using such an algorithm in route finding In this researchany user can access the ontology from any location and at anytime to conduct route finding Moreover being intelligent theubiquitous environment can receive and store the experiences(any data element of ubiquitous GIS) of any person to utilizewithin its services in the future

More innovative applications of spatial experiencemodeling such as ambulance firefighting and tourismservices can be subjects for further considerations -eprocess of route finding for this category is also affected bytraffic as emergency services can also become stuck in trafficlike other drivers because there are no dedicated lines on allroads Due to the limitation in length of road and a largenumber of vehicles in the road in traffic times it is hard forother vehicles to provide a route for traversing of the am-bulance and they still need to find a way to avoid traffic-erefore the experiences from expert technicians in am-bulances or firefighters who have been in real event con-ditions can be modeled and used in the future In additionthis study considers only two factors distance and time forevaluation Other criteria including fuel consumption can beconsidered in future

Data Availability

-e authors are ready to share the research data on request

Disclosure

Maryam Barzegar and Abolghasem Sadeghi-Niaraki are tobe considered as co-first authors

Conflicts of Interest

-e authors declare that they have no conflicts of interest

Acknowledgments

-is research was supported by theMSIT (Ministry of Scienceand ICT) Korea under the ITRC (Information Technology

0

01

02

03

041

2

3

4

5

6

7

8

9

10

DijkstraExperimental routes

Experimental traffic routesIdeal

Figure 17 Comparison of routes from the 3methods with the idealroute

Complexity 13

Research Center) support program (IITP-2019-2016-0-00312)supervised by the IITP (Institute for Information andCommunications Technology Planning and Evaluation)

References

[1] I P Tussyadiah and F J Zach ldquo-e role of geo-basedtechnology in place experiencesrdquo Annals of Tourism Researchvol 39 no 2 pp 780ndash800 2012

[2] G M Honti and J Abonyi ldquoA review of semantic sensortechnologies in internet of things architecturesrdquo Complexityvol 2019 Article ID 6473160 21 pages 2019

[3] U H Govindarajan A J Trappey and C V TrappeyldquoImmersive technology for human-centric cyberphysicalsystems in complex manufacturing processes a compre-hensive overview of the global patent profile using collectiveintelligencerdquo Complexity vol 2018 Article ID 428363417 pages 2018

[4] B K Foguem T Coudert C Beler and L GenesteldquoKnowledge formalization in experience feedback processesan ontology-based approachrdquo Computers in Industry vol 59no 7 pp 694ndash710 2008

[5] N Lasierra F Roldan A Alesanco and J Garcıa ldquoTowardsimproving usage and management of supplies in healthcarean ontology-based solution for sharing knowledgerdquo ExpertSystems with Applications vol 41 no 14 pp 6261ndash6273 2014

[6] M-H Abel ldquoKnowledge map-based web platform to facilitateorganizational learning return of experiencesrdquo Computers inHuman Behavior vol 51 pp 960ndash966 2015

[7] A Garcıa-Crespo J L Lopez-Cuadrado R Colomo-PalaciosI Gonzalez-Carrasco and B Ruiz-Mezcua ldquoSem-fit a se-mantic based expert system to provide recommendations inthe tourism domainrdquo Expert Systems with Applicationsvol 38 no 10 pp 13310ndash13319 2011

[8] D Mourtzis M Doukas and C Giannoulis ldquoAn inference-based knowledge reuse framework for historical product andproduction information retrievalrdquo Procedia CIRP vol 41pp 472ndash477 2016

[9] K Efthymiou K Sipsas D Mourtzis and G ChryssolourisldquoOn knowledge reuse for manufacturing systems design andplanning a semantic technology approachrdquo CIRP Journal ofManufacturing Science and Technology vol 8 pp 1ndash11 2015

[10] W L Mikos J C E Ferreira P E A Botura and L S FreitasldquoA system for distributed sharing and reuse of design andmanufacturing knowledge in the PFMEA domain using adescription logics-based ontologyrdquo Journal of ManufacturingSystems vol 30 no 3 pp 133ndash143 2011

[11] P P Ruiz B K Foguem and B Grabot ldquoGeneratingknowledge in maintenance from experience feedbackrdquoKnowledge-Based Systems vol 68 pp 4ndash20 2014

[12] S Moehrle and W Raskob ldquoStructuring and reusingknowledge from historical events for supporting nuclearemergency and remediation managementrdquo Engineering Ap-plications of Artificial Intelligence vol 46 pp 303ndash311 2015

[13] Q Meng Z Zhang X Wan and X Rong ldquoProperties ex-ploring and information mining in consumer communitynetwork a case of huawei pollen clubrdquo Complexity vol 2018Article ID 9470580 19 pages 2018

[14] R S Renu and GMocko ldquoComputing similarity of text-basedassembly processes for knowledge retrieval and reuserdquoJournal of Manufacturing Systems vol 39 pp 101ndash110 2016

[15] G E Modoni M Doukas W Terkaj M Sacco andD Mourtzis ldquoEnhancing factory data integration through thedevelopment of an ontology from the reference models reuse

to the semantic conversion of the legacy modelsrdquo In-ternational Journal of Computer Integrated Manufacturingvol 30 no 10 pp 1043ndash1059 2017

[16] E R Reyes S Negny G C Robles and J M Le LannldquoImprovement of online adaptation knowledge acquisitionand reuse in case-based reasoning application to processengineering designrdquo Engineering Applications of ArtificialIntelligence vol 41 pp 1ndash16 2015

[17] B Kamsu-Foguem and F H Abanda ldquoExperience modelingwith graphs encoded knowledge for construction industryrdquoComputers in Industry vol 70 pp 79ndash88 2015

[18] B Kamsu-Foguem F H Abanda M B Doumbouya andJ F Tchouanguem ldquoGraph-based ontology reasoning forformal verification of BREEAM rulesrdquo Cognitive SystemsResearch vol 55 pp 14ndash33 2019

[19] S H Othman and G Beydoun ldquoA metamodel-basedknowledge sharing system for disaster managementrdquo ExpertSystems with Applications vol 63 pp 49ndash65 2016

[20] D Xia X Lu H Li W Wang Y Li and Z Zhang ldquoAmapreduce-based parallel frequent pattern growth algorithmfor spatiotemporal association analysis of mobile trajectorybig datardquo Complexity vol 2018 Article ID 2818251 16 pages2018

[21] J Geng S Wang W Gan et al ldquoPromoting geospatial servicefrom information to knowledge with spatiotemporal se-manticsrdquo Complexity vol 2019 Article ID 9301420 14 pages2019

[22] A Nowak-Brzezinska ldquoEnhancing the efficiency of a decisionsupport system through the clustering of complex rule-basedknowledge bases and modification of the inference algo-rithmrdquo Complexity vol 2018 Article ID 2065491 14 pages2018

[23] E Maleki F Belkadi N Boli et al ldquoOntology-basedframework enabling smart product-service systems appli-cation of sensing systems for machine health monitoringrdquoIEEE Internet of ings Journal vol 5 no 6 pp 4496ndash45052018

[24] F H Abanda B Kamsu-Foguem and J H M TahldquoBIMmdashnew rules of measurement ontology for constructioncost estimationrdquo Engineering Science and Technology anInternational Journal vol 20 no 2 pp 443ndash459 2017

[25] B Kamsu-Foguem and P Tiako ldquoRisk information formal-isation with graphsrdquo Computers in Industry vol 85 pp 58ndash69 2017

[26] M Barzegar A Sadeghi-Niaraki and M Shakeri ldquoSpatialexperience based route finding using ontologiesrdquo ETRIJournal pp 1ndash11 2019

[27] E Camossi P Villa and L Mazzola ldquoSemantic-basedanomalous pattern discovery in moving object trajectoriesrdquo2013 httpsarxivorgabs13051946

[28] R Wannous J Malki A Bouju and C Vincent ldquoModellingmobile object activities based on trajectory ontology rulesconsidering spatial relationship rulesrdquo in Modeling Ap-proaches and Algorithms for Advanced Computer Applicationspp 249ndash258 Springer International Publishing BerlinGermany 2013

[29] Y Hu K Janowicz D Carral et al ldquoA geo-ontology designpattern for semantic trajectoriesrdquo in Proceedings of the In-ternational Conference on Spatial Information eorypp 438ndash456 Springer International Publishing ScarboroughUK September 2013

[30] M Baglioni J Macedo C Renso and M Wachowicz ldquoAnontology-based approach for the semantic modelling andreasoning on trajectoriesrdquo in Advances in Conceptual

14 Complexity

It is observable from the figures that travel timesobtained from the hierarchical experimental methodwhen only considering high-traffic conditions are lowesthowever the travel time of the shortest path is the highest-e mean travel time is 80 minutes for Dijkstrarsquos algo-rithm 61 minutes when only considering the high-trafficroutes and 775 minutes when considering all of theexperimental routes Although the routes are long whenconsidering high-traffic routes they are fast due to theselection of low-traffic routes in this case For examplewhen going to Tajrish Square from Valiasr SquareDijkstrarsquos algorithm suggests driving through ValiasrStreet which corresponds to the shortest but not fastestroute However the proposed ontology-based approachsuggests that the driver proceeds mainly along ModarresHighway which reduces the travel time by 35 minutes inhigh-traffic conditions

Another approach to evaluate the two proposedmethodsagainst Dijkstrarsquos algorithm is to find the ideal route amongthem In order to find the ideal route one can divide theroute length by its travel time Obviously when the routelengths for all routes or just two of them are equal the onewhich has the shortest travel time should be selected as theoptimum path On the other hand when the travel times areequal the one which has the shortest length should be se-lected as the optimum path Accordingly in eight cases (of10 cases) the ontology-based route finding considering onlyhigh-traffic routes (ETR) predominates In the other twocases the travel time was roughly equal however the routelengths obtained fromDijkstrarsquos algorithm had lower valuesTo determine the deviation of each method from the idealcase which is shown in Figure 17 the following formula isused

Fj travel length of method j

travel time of method j

j ETR ER Dikjestra

Fe ideal route fraction

Dj Fj minus Fe1113872 1113873

Fe

(3)

-erefore according to the evaluations above it can beclaimed that eliminating the experimental high-traffic routesin high-traffic conditions and utilizing the driversrsquo

02468

10121416

Mea

n of

rout

e len

gth

(km

)

Dijkstra Experimentalroutes

Method name

Experimentaltrafficroutes

Figure 13 Mean of route length for 3 methods

0

20

40

60

80

100

120

Trav

el ti

me (

min

)

1 2 3 4 5 6 7 8 9 10Route number

DijkstraExperimental routesExperimental traffic routes

Figure 14 Comparison of travel time for 10 evaluated routes of 3methods

002040608

112

1 2 3 4 5 6 7 8 9 10

ERDijkstraETRDijkstra

Figure 15 Comparison of travel time for the 2 modes of theproposed method with Dijkstrarsquos algorithm

Dijkstra020406080

Mea

n of

trav

el ti

me (

min

)

Experimentalroutes

Method name

Experimentaltrafficroutes

Figure 16 Mean of travel time for 3 methods

0

05

1

15

2

1 2 3 4 5 6 7 8 9 10

ERDijkstraETRDijkstra

Figure 12 Comparison of route length in 2 modes of the proposedalgorithm with Dijkstrarsquos algorithm

12 Complexity

experiences make the algorithm generate the optimum routethrough low-traffic parts of the network -is minimizestravel time which is the most important parameter in routefinding

6 Conclusions

Different people gain different spatial experiences Asthere are no systems to store these experiences theyevaporate over time-e present study provides the abilityto store spatial experiences and to make the existingsystems in ubiquitous GIS space intelligent In this paperan ontology-based route-finding algorithm in ubiquitousGIS space was designed and implemented using theroutes driven by the drivers of Tehran in high- and low-traffic conditions

In the proposed route-finding algorithm based on thedriversrsquo experiences ontology the number of times that eachsegment is passed by drivers is used to assign weights to thatsegment of the network First the user specifies the location thedestination and the time -en the algorithm determineswhether a subclass with the specified location and destinationexists in the ontology or not If the time chosen falls duringpeak times the existing paths in the traffic subclass are searchedthrough Otherwise the paths in the nontraffic subclass areexamined When a path exists the stored route is displayed tothe user and the algorithm terminates without needing toconduct route finding However when there is no corre-sponding path the algorithm determines whether the locationand destination exist in the experimental road network or notIf both exist in the network only the experimental is used toconstruct the graph and find the route (again peak time

determines the subclass to choosemdashtraffic or nontraffic) andthe algorithm terminates If the origin andor the destinationdoes not exist in the experimental road network the closestpoints of the network to them are found-en a rectangle witha diameter (R2R1 + 2radic2h) greater than the distance betweenthe location (or destination) and the point is extracted from themain road network -e main difference of this algorithm tothe previous experimental route-finding algorithms is the useof ontology and the construction of road networks onlyconsidering the low-traffic paths when there are high-trafficpaths in the network Since each newpath is stored as a separateOWL class the storage issues in the databases are resolved Inaddition the use of ontology causes the system to update everytime the algorithm runs and generates new paths Utilizing onlylow-traffic routes in high-traffic conditions allows the system tosuggest routes that are faster ensuring the optimum path basedon travel time is achieved According to our evaluations theproposed algorithm suggests routes that are longer but si-multaneously the fastest -is guarantees the appropriatenessof using such an algorithm in route finding In this researchany user can access the ontology from any location and at anytime to conduct route finding Moreover being intelligent theubiquitous environment can receive and store the experiences(any data element of ubiquitous GIS) of any person to utilizewithin its services in the future

More innovative applications of spatial experiencemodeling such as ambulance firefighting and tourismservices can be subjects for further considerations -eprocess of route finding for this category is also affected bytraffic as emergency services can also become stuck in trafficlike other drivers because there are no dedicated lines on allroads Due to the limitation in length of road and a largenumber of vehicles in the road in traffic times it is hard forother vehicles to provide a route for traversing of the am-bulance and they still need to find a way to avoid traffic-erefore the experiences from expert technicians in am-bulances or firefighters who have been in real event con-ditions can be modeled and used in the future In additionthis study considers only two factors distance and time forevaluation Other criteria including fuel consumption can beconsidered in future

Data Availability

-e authors are ready to share the research data on request

Disclosure

Maryam Barzegar and Abolghasem Sadeghi-Niaraki are tobe considered as co-first authors

Conflicts of Interest

-e authors declare that they have no conflicts of interest

Acknowledgments

-is research was supported by theMSIT (Ministry of Scienceand ICT) Korea under the ITRC (Information Technology

0

01

02

03

041

2

3

4

5

6

7

8

9

10

DijkstraExperimental routes

Experimental traffic routesIdeal

Figure 17 Comparison of routes from the 3methods with the idealroute

Complexity 13

Research Center) support program (IITP-2019-2016-0-00312)supervised by the IITP (Institute for Information andCommunications Technology Planning and Evaluation)

References

[1] I P Tussyadiah and F J Zach ldquo-e role of geo-basedtechnology in place experiencesrdquo Annals of Tourism Researchvol 39 no 2 pp 780ndash800 2012

[2] G M Honti and J Abonyi ldquoA review of semantic sensortechnologies in internet of things architecturesrdquo Complexityvol 2019 Article ID 6473160 21 pages 2019

[3] U H Govindarajan A J Trappey and C V TrappeyldquoImmersive technology for human-centric cyberphysicalsystems in complex manufacturing processes a compre-hensive overview of the global patent profile using collectiveintelligencerdquo Complexity vol 2018 Article ID 428363417 pages 2018

[4] B K Foguem T Coudert C Beler and L GenesteldquoKnowledge formalization in experience feedback processesan ontology-based approachrdquo Computers in Industry vol 59no 7 pp 694ndash710 2008

[5] N Lasierra F Roldan A Alesanco and J Garcıa ldquoTowardsimproving usage and management of supplies in healthcarean ontology-based solution for sharing knowledgerdquo ExpertSystems with Applications vol 41 no 14 pp 6261ndash6273 2014

[6] M-H Abel ldquoKnowledge map-based web platform to facilitateorganizational learning return of experiencesrdquo Computers inHuman Behavior vol 51 pp 960ndash966 2015

[7] A Garcıa-Crespo J L Lopez-Cuadrado R Colomo-PalaciosI Gonzalez-Carrasco and B Ruiz-Mezcua ldquoSem-fit a se-mantic based expert system to provide recommendations inthe tourism domainrdquo Expert Systems with Applicationsvol 38 no 10 pp 13310ndash13319 2011

[8] D Mourtzis M Doukas and C Giannoulis ldquoAn inference-based knowledge reuse framework for historical product andproduction information retrievalrdquo Procedia CIRP vol 41pp 472ndash477 2016

[9] K Efthymiou K Sipsas D Mourtzis and G ChryssolourisldquoOn knowledge reuse for manufacturing systems design andplanning a semantic technology approachrdquo CIRP Journal ofManufacturing Science and Technology vol 8 pp 1ndash11 2015

[10] W L Mikos J C E Ferreira P E A Botura and L S FreitasldquoA system for distributed sharing and reuse of design andmanufacturing knowledge in the PFMEA domain using adescription logics-based ontologyrdquo Journal of ManufacturingSystems vol 30 no 3 pp 133ndash143 2011

[11] P P Ruiz B K Foguem and B Grabot ldquoGeneratingknowledge in maintenance from experience feedbackrdquoKnowledge-Based Systems vol 68 pp 4ndash20 2014

[12] S Moehrle and W Raskob ldquoStructuring and reusingknowledge from historical events for supporting nuclearemergency and remediation managementrdquo Engineering Ap-plications of Artificial Intelligence vol 46 pp 303ndash311 2015

[13] Q Meng Z Zhang X Wan and X Rong ldquoProperties ex-ploring and information mining in consumer communitynetwork a case of huawei pollen clubrdquo Complexity vol 2018Article ID 9470580 19 pages 2018

[14] R S Renu and GMocko ldquoComputing similarity of text-basedassembly processes for knowledge retrieval and reuserdquoJournal of Manufacturing Systems vol 39 pp 101ndash110 2016

[15] G E Modoni M Doukas W Terkaj M Sacco andD Mourtzis ldquoEnhancing factory data integration through thedevelopment of an ontology from the reference models reuse

to the semantic conversion of the legacy modelsrdquo In-ternational Journal of Computer Integrated Manufacturingvol 30 no 10 pp 1043ndash1059 2017

[16] E R Reyes S Negny G C Robles and J M Le LannldquoImprovement of online adaptation knowledge acquisitionand reuse in case-based reasoning application to processengineering designrdquo Engineering Applications of ArtificialIntelligence vol 41 pp 1ndash16 2015

[17] B Kamsu-Foguem and F H Abanda ldquoExperience modelingwith graphs encoded knowledge for construction industryrdquoComputers in Industry vol 70 pp 79ndash88 2015

[18] B Kamsu-Foguem F H Abanda M B Doumbouya andJ F Tchouanguem ldquoGraph-based ontology reasoning forformal verification of BREEAM rulesrdquo Cognitive SystemsResearch vol 55 pp 14ndash33 2019

[19] S H Othman and G Beydoun ldquoA metamodel-basedknowledge sharing system for disaster managementrdquo ExpertSystems with Applications vol 63 pp 49ndash65 2016

[20] D Xia X Lu H Li W Wang Y Li and Z Zhang ldquoAmapreduce-based parallel frequent pattern growth algorithmfor spatiotemporal association analysis of mobile trajectorybig datardquo Complexity vol 2018 Article ID 2818251 16 pages2018

[21] J Geng S Wang W Gan et al ldquoPromoting geospatial servicefrom information to knowledge with spatiotemporal se-manticsrdquo Complexity vol 2019 Article ID 9301420 14 pages2019

[22] A Nowak-Brzezinska ldquoEnhancing the efficiency of a decisionsupport system through the clustering of complex rule-basedknowledge bases and modification of the inference algo-rithmrdquo Complexity vol 2018 Article ID 2065491 14 pages2018

[23] E Maleki F Belkadi N Boli et al ldquoOntology-basedframework enabling smart product-service systems appli-cation of sensing systems for machine health monitoringrdquoIEEE Internet of ings Journal vol 5 no 6 pp 4496ndash45052018

[24] F H Abanda B Kamsu-Foguem and J H M TahldquoBIMmdashnew rules of measurement ontology for constructioncost estimationrdquo Engineering Science and Technology anInternational Journal vol 20 no 2 pp 443ndash459 2017

[25] B Kamsu-Foguem and P Tiako ldquoRisk information formal-isation with graphsrdquo Computers in Industry vol 85 pp 58ndash69 2017

[26] M Barzegar A Sadeghi-Niaraki and M Shakeri ldquoSpatialexperience based route finding using ontologiesrdquo ETRIJournal pp 1ndash11 2019

[27] E Camossi P Villa and L Mazzola ldquoSemantic-basedanomalous pattern discovery in moving object trajectoriesrdquo2013 httpsarxivorgabs13051946

[28] R Wannous J Malki A Bouju and C Vincent ldquoModellingmobile object activities based on trajectory ontology rulesconsidering spatial relationship rulesrdquo in Modeling Ap-proaches and Algorithms for Advanced Computer Applicationspp 249ndash258 Springer International Publishing BerlinGermany 2013

[29] Y Hu K Janowicz D Carral et al ldquoA geo-ontology designpattern for semantic trajectoriesrdquo in Proceedings of the In-ternational Conference on Spatial Information eorypp 438ndash456 Springer International Publishing ScarboroughUK September 2013

[30] M Baglioni J Macedo C Renso and M Wachowicz ldquoAnontology-based approach for the semantic modelling andreasoning on trajectoriesrdquo in Advances in Conceptual

14 Complexity

experiences make the algorithm generate the optimum routethrough low-traffic parts of the network -is minimizestravel time which is the most important parameter in routefinding

6 Conclusions

Different people gain different spatial experiences Asthere are no systems to store these experiences theyevaporate over time-e present study provides the abilityto store spatial experiences and to make the existingsystems in ubiquitous GIS space intelligent In this paperan ontology-based route-finding algorithm in ubiquitousGIS space was designed and implemented using theroutes driven by the drivers of Tehran in high- and low-traffic conditions

In the proposed route-finding algorithm based on thedriversrsquo experiences ontology the number of times that eachsegment is passed by drivers is used to assign weights to thatsegment of the network First the user specifies the location thedestination and the time -en the algorithm determineswhether a subclass with the specified location and destinationexists in the ontology or not If the time chosen falls duringpeak times the existing paths in the traffic subclass are searchedthrough Otherwise the paths in the nontraffic subclass areexamined When a path exists the stored route is displayed tothe user and the algorithm terminates without needing toconduct route finding However when there is no corre-sponding path the algorithm determines whether the locationand destination exist in the experimental road network or notIf both exist in the network only the experimental is used toconstruct the graph and find the route (again peak time

determines the subclass to choosemdashtraffic or nontraffic) andthe algorithm terminates If the origin andor the destinationdoes not exist in the experimental road network the closestpoints of the network to them are found-en a rectangle witha diameter (R2R1 + 2radic2h) greater than the distance betweenthe location (or destination) and the point is extracted from themain road network -e main difference of this algorithm tothe previous experimental route-finding algorithms is the useof ontology and the construction of road networks onlyconsidering the low-traffic paths when there are high-trafficpaths in the network Since each newpath is stored as a separateOWL class the storage issues in the databases are resolved Inaddition the use of ontology causes the system to update everytime the algorithm runs and generates new paths Utilizing onlylow-traffic routes in high-traffic conditions allows the system tosuggest routes that are faster ensuring the optimum path basedon travel time is achieved According to our evaluations theproposed algorithm suggests routes that are longer but si-multaneously the fastest -is guarantees the appropriatenessof using such an algorithm in route finding In this researchany user can access the ontology from any location and at anytime to conduct route finding Moreover being intelligent theubiquitous environment can receive and store the experiences(any data element of ubiquitous GIS) of any person to utilizewithin its services in the future

More innovative applications of spatial experiencemodeling such as ambulance firefighting and tourismservices can be subjects for further considerations -eprocess of route finding for this category is also affected bytraffic as emergency services can also become stuck in trafficlike other drivers because there are no dedicated lines on allroads Due to the limitation in length of road and a largenumber of vehicles in the road in traffic times it is hard forother vehicles to provide a route for traversing of the am-bulance and they still need to find a way to avoid traffic-erefore the experiences from expert technicians in am-bulances or firefighters who have been in real event con-ditions can be modeled and used in the future In additionthis study considers only two factors distance and time forevaluation Other criteria including fuel consumption can beconsidered in future

Data Availability

-e authors are ready to share the research data on request

Disclosure

Maryam Barzegar and Abolghasem Sadeghi-Niaraki are tobe considered as co-first authors

Conflicts of Interest

-e authors declare that they have no conflicts of interest

Acknowledgments

-is research was supported by theMSIT (Ministry of Scienceand ICT) Korea under the ITRC (Information Technology

0

01

02

03

041

2

3

4

5

6

7

8

9

10

DijkstraExperimental routes

Experimental traffic routesIdeal

Figure 17 Comparison of routes from the 3methods with the idealroute

Complexity 13

Research Center) support program (IITP-2019-2016-0-00312)supervised by the IITP (Institute for Information andCommunications Technology Planning and Evaluation)

References

[1] I P Tussyadiah and F J Zach ldquo-e role of geo-basedtechnology in place experiencesrdquo Annals of Tourism Researchvol 39 no 2 pp 780ndash800 2012

[2] G M Honti and J Abonyi ldquoA review of semantic sensortechnologies in internet of things architecturesrdquo Complexityvol 2019 Article ID 6473160 21 pages 2019

[3] U H Govindarajan A J Trappey and C V TrappeyldquoImmersive technology for human-centric cyberphysicalsystems in complex manufacturing processes a compre-hensive overview of the global patent profile using collectiveintelligencerdquo Complexity vol 2018 Article ID 428363417 pages 2018

[4] B K Foguem T Coudert C Beler and L GenesteldquoKnowledge formalization in experience feedback processesan ontology-based approachrdquo Computers in Industry vol 59no 7 pp 694ndash710 2008

[5] N Lasierra F Roldan A Alesanco and J Garcıa ldquoTowardsimproving usage and management of supplies in healthcarean ontology-based solution for sharing knowledgerdquo ExpertSystems with Applications vol 41 no 14 pp 6261ndash6273 2014

[6] M-H Abel ldquoKnowledge map-based web platform to facilitateorganizational learning return of experiencesrdquo Computers inHuman Behavior vol 51 pp 960ndash966 2015

[7] A Garcıa-Crespo J L Lopez-Cuadrado R Colomo-PalaciosI Gonzalez-Carrasco and B Ruiz-Mezcua ldquoSem-fit a se-mantic based expert system to provide recommendations inthe tourism domainrdquo Expert Systems with Applicationsvol 38 no 10 pp 13310ndash13319 2011

[8] D Mourtzis M Doukas and C Giannoulis ldquoAn inference-based knowledge reuse framework for historical product andproduction information retrievalrdquo Procedia CIRP vol 41pp 472ndash477 2016

[9] K Efthymiou K Sipsas D Mourtzis and G ChryssolourisldquoOn knowledge reuse for manufacturing systems design andplanning a semantic technology approachrdquo CIRP Journal ofManufacturing Science and Technology vol 8 pp 1ndash11 2015

[10] W L Mikos J C E Ferreira P E A Botura and L S FreitasldquoA system for distributed sharing and reuse of design andmanufacturing knowledge in the PFMEA domain using adescription logics-based ontologyrdquo Journal of ManufacturingSystems vol 30 no 3 pp 133ndash143 2011

[11] P P Ruiz B K Foguem and B Grabot ldquoGeneratingknowledge in maintenance from experience feedbackrdquoKnowledge-Based Systems vol 68 pp 4ndash20 2014

[12] S Moehrle and W Raskob ldquoStructuring and reusingknowledge from historical events for supporting nuclearemergency and remediation managementrdquo Engineering Ap-plications of Artificial Intelligence vol 46 pp 303ndash311 2015

[13] Q Meng Z Zhang X Wan and X Rong ldquoProperties ex-ploring and information mining in consumer communitynetwork a case of huawei pollen clubrdquo Complexity vol 2018Article ID 9470580 19 pages 2018

[14] R S Renu and GMocko ldquoComputing similarity of text-basedassembly processes for knowledge retrieval and reuserdquoJournal of Manufacturing Systems vol 39 pp 101ndash110 2016

[15] G E Modoni M Doukas W Terkaj M Sacco andD Mourtzis ldquoEnhancing factory data integration through thedevelopment of an ontology from the reference models reuse

to the semantic conversion of the legacy modelsrdquo In-ternational Journal of Computer Integrated Manufacturingvol 30 no 10 pp 1043ndash1059 2017

[16] E R Reyes S Negny G C Robles and J M Le LannldquoImprovement of online adaptation knowledge acquisitionand reuse in case-based reasoning application to processengineering designrdquo Engineering Applications of ArtificialIntelligence vol 41 pp 1ndash16 2015

[17] B Kamsu-Foguem and F H Abanda ldquoExperience modelingwith graphs encoded knowledge for construction industryrdquoComputers in Industry vol 70 pp 79ndash88 2015

[18] B Kamsu-Foguem F H Abanda M B Doumbouya andJ F Tchouanguem ldquoGraph-based ontology reasoning forformal verification of BREEAM rulesrdquo Cognitive SystemsResearch vol 55 pp 14ndash33 2019

[19] S H Othman and G Beydoun ldquoA metamodel-basedknowledge sharing system for disaster managementrdquo ExpertSystems with Applications vol 63 pp 49ndash65 2016

[20] D Xia X Lu H Li W Wang Y Li and Z Zhang ldquoAmapreduce-based parallel frequent pattern growth algorithmfor spatiotemporal association analysis of mobile trajectorybig datardquo Complexity vol 2018 Article ID 2818251 16 pages2018

[21] J Geng S Wang W Gan et al ldquoPromoting geospatial servicefrom information to knowledge with spatiotemporal se-manticsrdquo Complexity vol 2019 Article ID 9301420 14 pages2019

[22] A Nowak-Brzezinska ldquoEnhancing the efficiency of a decisionsupport system through the clustering of complex rule-basedknowledge bases and modification of the inference algo-rithmrdquo Complexity vol 2018 Article ID 2065491 14 pages2018

[23] E Maleki F Belkadi N Boli et al ldquoOntology-basedframework enabling smart product-service systems appli-cation of sensing systems for machine health monitoringrdquoIEEE Internet of ings Journal vol 5 no 6 pp 4496ndash45052018

[24] F H Abanda B Kamsu-Foguem and J H M TahldquoBIMmdashnew rules of measurement ontology for constructioncost estimationrdquo Engineering Science and Technology anInternational Journal vol 20 no 2 pp 443ndash459 2017

[25] B Kamsu-Foguem and P Tiako ldquoRisk information formal-isation with graphsrdquo Computers in Industry vol 85 pp 58ndash69 2017

[26] M Barzegar A Sadeghi-Niaraki and M Shakeri ldquoSpatialexperience based route finding using ontologiesrdquo ETRIJournal pp 1ndash11 2019

[27] E Camossi P Villa and L Mazzola ldquoSemantic-basedanomalous pattern discovery in moving object trajectoriesrdquo2013 httpsarxivorgabs13051946

[28] R Wannous J Malki A Bouju and C Vincent ldquoModellingmobile object activities based on trajectory ontology rulesconsidering spatial relationship rulesrdquo in Modeling Ap-proaches and Algorithms for Advanced Computer Applicationspp 249ndash258 Springer International Publishing BerlinGermany 2013

[29] Y Hu K Janowicz D Carral et al ldquoA geo-ontology designpattern for semantic trajectoriesrdquo in Proceedings of the In-ternational Conference on Spatial Information eorypp 438ndash456 Springer International Publishing ScarboroughUK September 2013

[30] M Baglioni J Macedo C Renso and M Wachowicz ldquoAnontology-based approach for the semantic modelling andreasoning on trajectoriesrdquo in Advances in Conceptual

14 Complexity

Research Center) support program (IITP-2019-2016-0-00312)supervised by the IITP (Institute for Information andCommunications Technology Planning and Evaluation)

References

[1] I P Tussyadiah and F J Zach ldquo-e role of geo-basedtechnology in place experiencesrdquo Annals of Tourism Researchvol 39 no 2 pp 780ndash800 2012

[2] G M Honti and J Abonyi ldquoA review of semantic sensortechnologies in internet of things architecturesrdquo Complexityvol 2019 Article ID 6473160 21 pages 2019

[3] U H Govindarajan A J Trappey and C V TrappeyldquoImmersive technology for human-centric cyberphysicalsystems in complex manufacturing processes a compre-hensive overview of the global patent profile using collectiveintelligencerdquo Complexity vol 2018 Article ID 428363417 pages 2018

[4] B K Foguem T Coudert C Beler and L GenesteldquoKnowledge formalization in experience feedback processesan ontology-based approachrdquo Computers in Industry vol 59no 7 pp 694ndash710 2008

[5] N Lasierra F Roldan A Alesanco and J Garcıa ldquoTowardsimproving usage and management of supplies in healthcarean ontology-based solution for sharing knowledgerdquo ExpertSystems with Applications vol 41 no 14 pp 6261ndash6273 2014

[6] M-H Abel ldquoKnowledge map-based web platform to facilitateorganizational learning return of experiencesrdquo Computers inHuman Behavior vol 51 pp 960ndash966 2015

[7] A Garcıa-Crespo J L Lopez-Cuadrado R Colomo-PalaciosI Gonzalez-Carrasco and B Ruiz-Mezcua ldquoSem-fit a se-mantic based expert system to provide recommendations inthe tourism domainrdquo Expert Systems with Applicationsvol 38 no 10 pp 13310ndash13319 2011

[8] D Mourtzis M Doukas and C Giannoulis ldquoAn inference-based knowledge reuse framework for historical product andproduction information retrievalrdquo Procedia CIRP vol 41pp 472ndash477 2016

[9] K Efthymiou K Sipsas D Mourtzis and G ChryssolourisldquoOn knowledge reuse for manufacturing systems design andplanning a semantic technology approachrdquo CIRP Journal ofManufacturing Science and Technology vol 8 pp 1ndash11 2015

[10] W L Mikos J C E Ferreira P E A Botura and L S FreitasldquoA system for distributed sharing and reuse of design andmanufacturing knowledge in the PFMEA domain using adescription logics-based ontologyrdquo Journal of ManufacturingSystems vol 30 no 3 pp 133ndash143 2011

[11] P P Ruiz B K Foguem and B Grabot ldquoGeneratingknowledge in maintenance from experience feedbackrdquoKnowledge-Based Systems vol 68 pp 4ndash20 2014

[12] S Moehrle and W Raskob ldquoStructuring and reusingknowledge from historical events for supporting nuclearemergency and remediation managementrdquo Engineering Ap-plications of Artificial Intelligence vol 46 pp 303ndash311 2015

[13] Q Meng Z Zhang X Wan and X Rong ldquoProperties ex-ploring and information mining in consumer communitynetwork a case of huawei pollen clubrdquo Complexity vol 2018Article ID 9470580 19 pages 2018

[14] R S Renu and GMocko ldquoComputing similarity of text-basedassembly processes for knowledge retrieval and reuserdquoJournal of Manufacturing Systems vol 39 pp 101ndash110 2016

[15] G E Modoni M Doukas W Terkaj M Sacco andD Mourtzis ldquoEnhancing factory data integration through thedevelopment of an ontology from the reference models reuse

to the semantic conversion of the legacy modelsrdquo In-ternational Journal of Computer Integrated Manufacturingvol 30 no 10 pp 1043ndash1059 2017

[16] E R Reyes S Negny G C Robles and J M Le LannldquoImprovement of online adaptation knowledge acquisitionand reuse in case-based reasoning application to processengineering designrdquo Engineering Applications of ArtificialIntelligence vol 41 pp 1ndash16 2015

[17] B Kamsu-Foguem and F H Abanda ldquoExperience modelingwith graphs encoded knowledge for construction industryrdquoComputers in Industry vol 70 pp 79ndash88 2015

[18] B Kamsu-Foguem F H Abanda M B Doumbouya andJ F Tchouanguem ldquoGraph-based ontology reasoning forformal verification of BREEAM rulesrdquo Cognitive SystemsResearch vol 55 pp 14ndash33 2019

[19] S H Othman and G Beydoun ldquoA metamodel-basedknowledge sharing system for disaster managementrdquo ExpertSystems with Applications vol 63 pp 49ndash65 2016

[20] D Xia X Lu H Li W Wang Y Li and Z Zhang ldquoAmapreduce-based parallel frequent pattern growth algorithmfor spatiotemporal association analysis of mobile trajectorybig datardquo Complexity vol 2018 Article ID 2818251 16 pages2018

[21] J Geng S Wang W Gan et al ldquoPromoting geospatial servicefrom information to knowledge with spatiotemporal se-manticsrdquo Complexity vol 2019 Article ID 9301420 14 pages2019

[22] A Nowak-Brzezinska ldquoEnhancing the efficiency of a decisionsupport system through the clustering of complex rule-basedknowledge bases and modification of the inference algo-rithmrdquo Complexity vol 2018 Article ID 2065491 14 pages2018

[23] E Maleki F Belkadi N Boli et al ldquoOntology-basedframework enabling smart product-service systems appli-cation of sensing systems for machine health monitoringrdquoIEEE Internet of ings Journal vol 5 no 6 pp 4496ndash45052018

[24] F H Abanda B Kamsu-Foguem and J H M TahldquoBIMmdashnew rules of measurement ontology for constructioncost estimationrdquo Engineering Science and Technology anInternational Journal vol 20 no 2 pp 443ndash459 2017

[25] B Kamsu-Foguem and P Tiako ldquoRisk information formal-isation with graphsrdquo Computers in Industry vol 85 pp 58ndash69 2017

[26] M Barzegar A Sadeghi-Niaraki and M Shakeri ldquoSpatialexperience based route finding using ontologiesrdquo ETRIJournal pp 1ndash11 2019

[27] E Camossi P Villa and L Mazzola ldquoSemantic-basedanomalous pattern discovery in moving object trajectoriesrdquo2013 httpsarxivorgabs13051946

[28] R Wannous J Malki A Bouju and C Vincent ldquoModellingmobile object activities based on trajectory ontology rulesconsidering spatial relationship rulesrdquo in Modeling Ap-proaches and Algorithms for Advanced Computer Applicationspp 249ndash258 Springer International Publishing BerlinGermany 2013

[29] Y Hu K Janowicz D Carral et al ldquoA geo-ontology designpattern for semantic trajectoriesrdquo in Proceedings of the In-ternational Conference on Spatial Information eorypp 438ndash456 Springer International Publishing ScarboroughUK September 2013

[30] M Baglioni J Macedo C Renso and M Wachowicz ldquoAnontology-based approach for the semantic modelling andreasoning on trajectoriesrdquo in Advances in Conceptual

14 Complexity

ModelingmdashChallenges and Opportunities pp 344ndash353Springer Berlin Germany 2008

[31] U Durak H Oǧuztuzun and S K Ider ldquoOntology-baseddomain engineering for trajectory simulation reuserdquo In-ternational Journal of Software Engineering and KnowledgeEngineering vol 19 no 8 pp 1109ndash1129 2009

[32] T Malgundkar M Rao and S S Mantha ldquoGIS driven urbantraffic analysis based on ontologyrdquo International Journal ofManaging Information Technology vol 4 no 1 pp 15ndash232012

[33] A Sadeghi-Niaraki A Rajabifard K Kim and J Seo ldquoOn-tology based SDI to facilitate spatially enabled societyrdquo inProceedings of GSDI 12 World Conference pp 19ndash22 Sin-gapore October 2010

[34] A S Niaraki and K Kim ldquoOntology based personalized routeplanning system using a multi-criteria decision making ap-proachrdquo Expert Systems with Applications vol 36 no 2pp 2250ndash2259 2009

[35] S Saeedi N El-Sheimy M Malek and N Samani ldquoAnontology based context modeling approach for mobile touringand navigation systemrdquo in Proceedings of the the 2010 Ca-nadian Geomatics Conference and Symposium of CommissionI ISPRS Convergence in GeomaticsndashShaping Canadarsquos Com-petitive Landscape pp 15ndash18 Calgary Canada June 2010

[36] M Effati and A Sadeghi-Niaraki ldquoA semantic-based classi-fication and regression tree approach for modelling complexspatial rules in motor vehicle crashes domainrdquo Wiley In-terdisciplinary Reviews Data Mining and Knowledge Dis-covery vol 5 no 4 pp 181ndash194 2015

[37] J M Czerniak W Dobrosielski H Zarzycki andŁ Apiecionek ldquoA proposal of the new owlANT method fordetermining the distance between terms in ontologyrdquo inIntelligent Systemsrsquo 2014 pp 235ndash246 Springer ChamSwitzerland 2015

[38] Q Li Z Zeng T Zhang J Li and Z Wu ldquoPath-findingthrough flexible hierarchical road networks an experientialapproach using taxi trajectory datardquo International Journal ofApplied Earth Observation and Geoinformation vol 13 no 1pp 110ndash119 2011

[39] H Ji-hua H Ze and D Jun ldquoA hierarchical path planningmethod using the experience of taxi driversrdquo ProcediamdashSocialand Behavioral Sciences vol 96 pp 1898ndash1909 2013

[40] B D Ziebart A L Maas A K Dey and J A BagnellldquoNavigate like a cabbie probabilistic reasoning from observedcontext-aware behaviorrdquo in Proceedings of the 10th In-ternational Conference on Ubiquitous Computing pp 322ndash331 ACM Seoul South Korea September 2008

[41] J Yuan Y Zheng C Zhang et al ldquoT-drive driving directionsbased on taxi trajectoriesrdquo in Proceedings of the 18th SIG-SPATIAL International Conference on Advances in GeographicInformation Systems pp 99ndash108 ACM San Jose CA USANovember 2010

[42] C Chen D Zhang P S Castro N Li L Sun and S Li ldquoReal-time detection of anomalous taxi trajectories from GPStracesrdquo in Proceedings of the International Conference onMobile and Ubiquitous Systems Computing Networking andServices pp 63ndash74 Springer Berlin Heidelberg CopenhagenDenmark December 2011

[43] D Chu D A Sheets Y Zhao et al ldquoVisualizing hiddenthemes of taxi movement with semantic transformationrdquo inProceedings of the 2014 IEEE Pacific Visualization Symposiumpp 137ndash144 IEEE Yokohama Japan March 2014

[44] H Liu L Y Wei Y Zheng M Schneider and W C PengldquoRoute discovery from mining uncertain trajectoriesrdquo in

Proceedings of the 2011 IEEE 11th International Conference onData Mining Workshops pp 1239ndash1242 IEEE VancouverBC Canada December 2011

[45] X Liu L Gong Y Gong and Y Liu ldquoRevealing travelpatterns and city structure with taxi trip datardquo Journal ofTransport Geography vol 43 pp 78ndash90 2015

[46] M Rahmani and H N Koutsopoulos ldquoPath inference fromsparse floating car data for urban networksrdquo TransportationResearch Part C Emerging Technologies vol 30 pp 41ndash54 2013

[47] Y Zheng Q Li Y Chen X Xie and W Y Ma ldquoUn-derstandingmobility based on GPS datardquo in Proceedings of the10th International Conference on Ubiquitous Computingpp 312ndash321 ACM Seoul South Korea September 2008

[48] S Hasani A Sadeghi-Niaraki and M Jelokhani-NiarakildquoSpatial data integration using ontology-based approachrdquoISPRSmdashInternational Archives of the Photogrammetry Re-mote Sensing and Spatial Information Sciences vol XL-1-W5no 1 pp 293ndash296 2015

[49] P Sureephong N Chakpitak Y Ouzrout and A Bouras ldquoAnontology-based knowledge management system for industryclustersrdquo in Global Design to Gain a Competitive Edgepp 333ndash342 Springer London UK 2008

[50] M H Rashidan and I A Musliman ldquoGeoPackage as futureubiquitous GIS data format a reviewrdquo Jurnal Teknologivol 73 no 5 2015

[51] T J Kim and SG Jang ldquoUbiquitous geographic informationrdquo inSpringer Handbook of Geographic Information pp 369ndash378Springer Berlin Heidelberg Berlin Germany 2011

[52] S Mathew ldquoUbiquitous computing-A technological impactfor an intelligent systemrdquo International Journal of ComputerScience and Electronics Engineering (IJCSEE) vol 1 no 5pp 595ndash598 2013

[53] M Barzegar A Sadeghi-Niaraki M Shakeri and S-M Choi ldquoAcontext-aware route finding algorithm for self-driving touristsusing ontologyrdquo Electronics vol 8 no 7 808 pages 2019

[54] T Tran H Lewen and P Haase ldquoOn the role and application ofontologies in information systemsrdquo in Proceedings of the 2007IEEE International Conference on Research Innovation and Visionfor the Future (RIVF) pp 14ndash21 Hanoi Vietnam March 2007

[55] T Dillon E Chang M Hadzic and P WongthongthamldquoDifferentiating conceptual modelling from data modellingknowledge modelling and ontology modelling and a notationfor ontology modellingrdquo in Proceedings of the Fifth on Asia-Pacific Conference on Conceptual Modelling APCCMrsquo 08pp 7ndash17 Australian Computer Society Inc DarlinghurstAustralia May 2008

[56] M Breu and Y Ding ldquoModelling the world databases andontologiesrdquo in Whitepaper by IFI Institute of ComputerScience University of Innsbruck Innsbruck Austria 2004

[57] C Martinez-Cruz I J Blanco and M A Vila ldquoOntologiesversus relational databases are they so different A com-parisonrdquo Artificial Intelligence Review vol 38 no 4pp 271ndash290 2012

[58] S H Permana K Y Bintoro B Arifitama and A SyahputraldquoComparative analysis of pathfinding algorithms Alowast Dijkstraand BFS on maze runner gamerdquo IJISTECH (InternationalJournal of Information System and Technology) vol 1 no 21 pages 2018

Complexity 15

Hindawiwwwhindawicom Volume 2018

MathematicsJournal of

Hindawiwwwhindawicom Volume 2018

Mathematical Problems in Engineering

Applied MathematicsJournal of

Hindawiwwwhindawicom Volume 2018

Probability and StatisticsHindawiwwwhindawicom Volume 2018

Journal of

Hindawiwwwhindawicom Volume 2018

Mathematical PhysicsAdvances in

Complex AnalysisJournal of

Hindawiwwwhindawicom Volume 2018

OptimizationJournal of

Hindawiwwwhindawicom Volume 2018

Hindawiwwwhindawicom Volume 2018

Engineering Mathematics

International Journal of

Hindawiwwwhindawicom Volume 2018

Operations ResearchAdvances in

Journal of

Hindawiwwwhindawicom Volume 2018

Function SpacesAbstract and Applied AnalysisHindawiwwwhindawicom Volume 2018

International Journal of Mathematics and Mathematical Sciences

Hindawiwwwhindawicom Volume 2018

Hindawi Publishing Corporation httpwwwhindawicom Volume 2013Hindawiwwwhindawicom

The Scientific World Journal

Volume 2018

Hindawiwwwhindawicom Volume 2018Volume 2018

Numerical AnalysisNumerical AnalysisNumerical AnalysisNumerical AnalysisNumerical AnalysisNumerical AnalysisNumerical AnalysisNumerical AnalysisNumerical AnalysisNumerical AnalysisNumerical AnalysisNumerical AnalysisAdvances inAdvances in Discrete Dynamics in

Nature and SocietyHindawiwwwhindawicom Volume 2018

Hindawiwwwhindawicom

Dierential EquationsInternational Journal of

Volume 2018

Hindawiwwwhindawicom Volume 2018

Decision SciencesAdvances in

Hindawiwwwhindawicom Volume 2018

AnalysisInternational Journal of

Hindawiwwwhindawicom Volume 2018

Stochastic AnalysisInternational Journal of

Submit your manuscripts atwwwhindawicom

Hindawiwwwhindawicom Volume 2018

MathematicsJournal of

Hindawiwwwhindawicom Volume 2018

Mathematical Problems in Engineering

Applied MathematicsJournal of

Hindawiwwwhindawicom Volume 2018

Probability and StatisticsHindawiwwwhindawicom Volume 2018

Journal of

Hindawiwwwhindawicom Volume 2018

Mathematical PhysicsAdvances in

Complex AnalysisJournal of

Hindawiwwwhindawicom Volume 2018

OptimizationJournal of

Hindawiwwwhindawicom Volume 2018

Hindawiwwwhindawicom Volume 2018

Engineering Mathematics

International Journal of

Hindawiwwwhindawicom Volume 2018

Operations ResearchAdvances in

Journal of

Hindawiwwwhindawicom Volume 2018

Function SpacesAbstract and Applied AnalysisHindawiwwwhindawicom Volume 2018

International Journal of Mathematics and Mathematical Sciences

Hindawiwwwhindawicom Volume 2018

Hindawi Publishing Corporation httpwwwhindawicom Volume 2013Hindawiwwwhindawicom

The Scientific World Journal

Volume 2018

Hindawiwwwhindawicom Volume 2018Volume 2018

Numerical AnalysisNumerical AnalysisNumerical AnalysisNumerical AnalysisNumerical AnalysisNumerical AnalysisNumerical AnalysisNumerical AnalysisNumerical AnalysisNumerical AnalysisNumerical AnalysisNumerical AnalysisAdvances inAdvances in Discrete Dynamics in

Nature and SocietyHindawiwwwhindawicom Volume 2018

Hindawiwwwhindawicom

Dierential EquationsInternational Journal of

Volume 2018

Hindawiwwwhindawicom Volume 2018

Decision SciencesAdvances in

Hindawiwwwhindawicom Volume 2018

AnalysisInternational Journal of

Hindawiwwwhindawicom Volume 2018

Stochastic AnalysisInternational Journal of

Submit your manuscripts atwwwhindawicom