a study on development of cognitive support features in recent ontology visualization tools

Artif Intell Rev (2014) 41:595623DOI 10.1007/s10462-012-9326-2

A study on development of cognitive support featuresin recent ontology visualization tools

Sivakumar Ramakrishnan Arivoli Vijayan

Published online: 6 March 2012 Springer Science+Business Media B.V. 2012

Abstract Ontologies are currently emerging as representation techniques for overlappingcompatibility context domains. The continuing need for more effective information retrievalhas lead to the creation of the notions of the semantic web and personalized information man-agement. Subsequently, the need for effective ontology visualization for design, managementand browsing has arisen. Several ontology visualization tools have come out to strengthenthe users cognitive support. The primary goal of this paper is to present a survey on recentlyimplemented ontology visualization tools and their contributions in the enrichment of userscognitive support. This work also presents the preliminary results of an evaluation of threevisualization tools to determine the suitability of eachmethod for end user applications whereontologies are used as browsing aids.

Keywords Cognitive support Ontology visualization Semantic web Knowledge retrieval Reasoner

1 Introduction

Knowledge capture and knowledge representation are the hot emerging areas of research inthe discipline of information technology. Ontology development and semantic web play thevital role to support the research in knowledge representations. In recent years, number ofontology tools have been designed and implemented with the support of visualization. Thearea of cognitive assistance much requires visualization techniques for its improvement inperformance.

S. Ramakrishnan (B) A. VijayanDepartment of Computer Science, A.V.V.M Sri Pushpam College of Bharathidasan University,Thanjavur, Tamil Nadu, Indiae-mail: [email protected]

A. Vijayane-mail: [email protected]

123

596 S. Ramakrishnan, A. Vijayan

An ontology is a conceptualization of a domain into machine readable format (Guarinoand Giaretta 1995). Ontologies are becoming increasingly popular modelling schemas forknowledge management services and applications. Focus on developing tools to graphicallyvisualise ontologies is rising to aid their assessment and analysis. Graph visualisation helpsto browse and comprehend the structure of ontologies. Protg (Noy and McGuiness 2001)is one of the most widely used ontology development tools that were developed at StanfordUniversity. Protg provides an intuitive editor for ontologies and has extensions for ontol-ogy visualization, project management, software engineering and other modelling tasks. Anontology, according to the definition in Noy et al. (2001) is a formal explicit descriptionof a domain, consisting of classes, which are the concepts found in the domain. Classesare organized in a specialization/generalization hierarchy through is-a (or inheritance) links,where each class is allowed to have zero, one or multiple parent classes. Each class has prop-erties (or slots) describing various features of the modelled class. Slots are typed, and allowedtypes are either simple types (strings, numbers, booleans or enumerations) or instances ofother classes (references); restriction on the value ranges of slots (e.g. integers from 1 to10) may also be defined. Finally, instantiation may be applied to classes to produce itemscorresponding to individual objects in the domain of discourse (instances). Each instance hasa concrete value for each property of the class it belongs to. Classes, together with instancesare said to constitute the knowledge base. From the definition above, it is evident that the taskof visualizing the full set of ontology features is not an easy one. The properties of ontologyare summarized as follows:

Hierarchy. A type of organization that, like a tree, branches into more specific units, eachof which is owned by the higher-level unit immediately above.

Properties representation. More than a hierarchy, as it concepts are described by usingrestrictions on properties.

Level of detail. Possibility to choose till which level an ontology to be provided. History. The concepts that were chosen in the previous steps. Filtering. Ontologies could contain hundreds of properties. The user can be interested

in only the subset of the ontology, based on the central concept and the properties of theusers choice.

Multiple geometrical views. The representation of the graph in different geometricalmodels to better understand the structure of ontology.

Zoom semantic/geometric. To see more or less details during ontology exploration. Withthe geometric zoom the visualized object is scaled when the user zooms in/out. Thesemantic zoom provides the possibility to see more/less details of the object by zoomingin/out.

A number of ontology visualizations (Boinski et al. 2009;Benjamin et al. 2011) exist that havebeen embedded in ontology management tools (e.g. http://protege.stanford.edu/ and http://kaon.semanticweb.org/) and are used as information retrieval aids in applications that useontologies (Katifori et al. 2007). Evaluations of ontology visualization effectiveness, how-ever, are up to this point scarce: (Kobsa 2004) presents some user experiments focused ontree visualization systems, whereas (Katifori et al. 2006) reports on preliminary results froma user study involving four visualization methods. They are indented list, node-link and tree,zoomable, and focus+context. A number of visualisation techniques have been described overthe years, such as spanning tree layouts, tree-maps (Johnson and Shneiderman 1991) fisheyeviews (Furnas 1986), hyperbolic (Lamping et al. 1995) and 3D hyperbolic layouts (Munzner1997), aiming to help comprehend and analyse complex information structures. Preferenceof visualisationmodels vary according to users needs and query context (Graham et al. 2000).

123

http://protege.stanford.edu/http://kaon.semanticweb.org/http://kaon.semanticweb.org/

Recent ontology visualization tools 597

It is also dependant on the type and extent of the visualised network. Using a combinationof integrated visualisations of various types has shown to be sometimes beneficial (Northand Shneiderman 2000; Risden et al. 2000). Complex networks of multi dimensional hier-archies and arbitrary relations are becoming common characteristics of current ontologies.Tools that discriminate against some of these features, for example by supporting spanningtrees or hierarchical relations only, might not be appropriate for comprehensive ontologyvisualisation. Ontologies, together with their Knowledge Bases (KBs), could grow into verylarge information networks, especially if aimed at providing scalable services for the Seman-tic Web (Berners-Lee et al. 2001). Visualising large networks has always been challenging.Wills (1997), Herman et al. (2000) Surveyed a wide range of visualisation techniques andconcluded that all existing algorithms have a size limit after which they cannot cope (North1995) and (Herman et al. 2000) stressed the importance of reducing the visualised graphsinto smaller sized sub graphs that users can browse to visualise other parts of the network.Ontologies are semantically rich by definition. Ontology visualizes should therefore turnsome of these semantics more explicit (Wills 1997). Spring-layout algorithms (Eades 1984)are example techniques that display semantically similar nodes closer to each other. Suchlayouts could help users to quickly recognise dense areas and interrelated objects in theirontologies and KBs. In this paper we conduct a survey on identifying developments of cog-nitive support in recently implemented ontology visualization tools and present a summaryof various features and capabilities of those tools. The purpose of this work is to assist theresearchers of ontologies to understand more about this domain and to extend their researchactivities in new directions.

Information visualization (Boinski et al. 2010) represents information in a manner whichaids in communication and facilitates understanding and exploration. The view of toolsrequires the user to first understand what the view is attempting to show. Most innovativevisualizations suffer from lack of industry acceptance. This lack of adoption restricts mean-ingful evaluation of the ideas in the tool. With this concept of simplicity in mind, we need tomake conscious and deliberate efforts to focus in identifying a tool which will focus on thekey goals of communication and understanding, leveraging the techniques that best facilitatethose goals. Hence the next purpose of this paper is to evaluate the degree of cognitive sup-port offered in selected three different ontology visualization tools with the help of end usergroups.

The remaining part of this paper is organized as follows: Sect. 2 gives the introduction toProtg. Next Sect. 3 gives an overview of ontology visualization methods used in ontologyvisualization development. A survey on advances of recent ontology visualization tools ispresented in Sect. 4. Followed which, Sect. 5 comes out with evaluation of selected tools todetermine their effectiveness in cognitive support. Section 6 concluded with future work.

2 About Protg

Protg-2000 (Musen et al. 2000) is a very popular knowledge-modeling tool developedat Stanford University. Ontologies and knowledge-bases can be edited interactively withinProtg and accessed with a graphical user interface and Java API. Protg can be extendedwith pluggable components to add new functionalities and services. There exists an increasingnumber of plug-ins offering a variety of additional features, such as extra ontology manage-ment tools, multimedia support, querying and reasoning engines, problem solving methods,etc. Protg implements a rich set of knowledge-modeling structures and actions that supportthe creation, visualization, and manipulation of ontologies in various representation formats.

123


Protg gives support for building the ontologies that are frame-based, in accordance withthe Open Knowledge Base Connectivity protocol (OKBC). The extended version of framebased system was introduced in 2003 to support OWL with an advantage of semantic webversion. There are various forms such as RDF (Resource Description Framework), OWL(Ontology Web Language) and XML Schema in which protg ontology can be exported.The following are various plug-ins available in protg.

JSaveTo generate java class definition stubs for protg classes (http://protege.stanford.edu/plugins/jsave/).Protg Web BrowserAllows users to share protg ontologies over the internet.WordNet plug-inAn interface toWordNet knowledge base (http://protege.stanford.edu/plug-ins/wordnettab/wordnet_tab.html).XML SchemaTransforms a protg knowledge into XML (http://faculty.washington.edu/gennari/Protege-plugins/XMLBackend/XMLBackend.html).UML Plug-inGenerates an XML schema file describing protg knowledge model andan XML file (http://protege.stanford.edu/plug-ins/uml/).DataGenieAllows reading and creating knowledge model from RDBMS using JDBC(http://faculty.washington.edu/gennari/Protege-plugins/DataGenie/index.html).DocgenTocreate reports describingontologies (http://protege-docgen.sourceforge.net/).JessTabProvides a Jess console window to interact with Jess while running protg(http://www.ida.liu.se/~her/JessTab/).AlgernonPerforms forward and backward inference of frame based knowledge bases(http:/algernonj.sourceforge.net/doc/algernon-protege.html).PROMTAllows to manage multiple ontologies with in protg.OWL-S EditorAllows loading, creating, managing and visualizing OWL-S services(http://owlseditor.semwebcentral.org/).

3 Ontology visualization methods

This section gives an overview of most widely used ontology visualization methods.

3.1 The indented list method

The indented list methods represent the taxonomy of the ontology following the file systemexplorer-tree view. These methods are intuitive and simple to implement, representing is-ainheritance relationships through the indented list paradigm,with subclasses appearing belowtheir super classes and indented to the right. Users may navigate within the class hierarchyand expand or retract branches; when a class (or multiple classes) is (are) selected in thehierarchy pane, the corresponding instances are shown in the Instance browser (Katiforiet al. 2006; Nadia et al. 2009). The following Fig. 1 is example of the indented list method.

3.2 Node link and tree

The node-link and tree methods represent another approach frequently used for ontologyvisualization. The ontology is represented as a set of interconnected nodes. They offer agood overview of the hierarchy and connections, but may produce cluttered displays whenused to visualize more than hundred of nodes. The Fig. 2 shows node-link and tree method.

123

http://protege.stanford.edu/plugins/jsave/http://protege.stanford.edu/plugins/jsave/http://protege.stanford.edu/plug-ins/wordnettab/wordnet_tab.htmlhttp://protege.stanford.edu/plug-ins/wordnettab/wordnet_tab.htmlhttp://faculty.washington.edu/gennari/Protege-plugins/XMLBackend/XMLBackend.htmlhttp://faculty.washington.edu/gennari/Protege-plugins/XMLBackend/XMLBackend.htmlhttp://protege.stanford.edu/plug-ins/uml/http://faculty.washington.edu/gennari/Protege-plugins/DataGenie/index.htmlhttp://protege-docgen.sourceforge.net/http://www.ida.liu.se/~her/JessTab/http:/algernonj.sourceforge.net/doc/algernon-protege.htmlhttp://owlseditor.semwebcentral.org/


Fig. 1 Indented list method

Fig. 2 Node-link and tree method

3.3 Zoomable method

The zoomable methods present the nodes in the lower levels of the hierarchy nested insidetheir parents and with smaller size. The user may zoom-in to the child nodes to enlarge them.

123


Fig. 3 Zoomable method

Node nesting is also used for instance-of relationships, thus a class node contains both itssubclasses and its instances; the user may distinguish between the class-type and instance-type nodes through their color. These methods seem to be successful for browsing to locatespecific nodes. The Fig. 3 is example of zoomable method.

3.4 Focus+context method

The focus+context methods present the node on the focus in the centre and the connectednodes around it, usually reduced in size. According to this technique nodes (classes) repelone another, whereas the edges (links) attract them, thus nodes that are semantically similarare placed closed to one another. This technique is effective to provide global overviews, tofocus on specific nodes, and for quick browsing. The Fig. 4 shows focus+context method.

4 Advances in cognitive support:recent ontology visualization tools

This section surveys the recently implemented ontology visualization tools with a scope ofextended features that enhance users cognitive support.

4.1 SOVA (2011)

It stands for Simple Ontology Visualization API (Application Program Interface). It is aProtg plug-in to full ontology visualization. The users can view all ontology elements suchas classes, individuals, properties, anonymous classes and relations between these objects.This plug-in is an extension of an OWLAPIontology visualization library. It offers twotypes of visualization such as full ontology and hierarchy of classes and individuals. In firsttype, there are two choices under arrangement algorithm option. They are fore directedLay-out and RadialTreeLayout algorithms. The first uses a gravity algorithm to arrange elements

123


Fig. 4 Focus+Context method

while the second one puts ontology elements in circles around currently choosen elementaccording to distance between these elements. Next, in classified hierarchy of classes andindividuals representation, it uses pellet for classification process. This plug-in is designedfor application in Protg-OWL 4.0, Protg-OWL 4.1 and Protg-OWL 4.2. It providestwo main options such as PGETI SOVAHierarchy tree Vis (reset, show full tree and saveimage) and PGETI SOVAVisualization (play/stop, reset, options-filter and save image).

4.2 Snow OWL (2011)

This is a freeware multiplatform browser and authoring tool integrating Protg. It is builtfor healthcare application. Its dependency is Protg-OWL 4.1. Snow OWL is built upon theEclipse platform, which allows one to easily integrate Eclipse-based application architectureswithin a single tool. An example of this integration is Stanfords Protg Ontology Editor,which allows one to use any Protg-compatible description logic classifier. Snow Owl alsoseamlessly integrates with informationmodel tooling, such as theHL7 StaticModel Designerand the Logical InformationModel tooling to support domainmodel design, ontology author-ing, and the binding between the two. Snow Owl can operate in two modes; in stand-alonemode supporting a single user or in collaborative mode with multiple users connected to asingle server authoring the same terminology simultaneously in a distributed fashion.Key features include:

Browsing:SnowOwl supports browsing installed terminologies by browsing andfilteringthe concept hierarchy.

Searching: A search window allows more advanced searching options than present inthe individual terminology trees. It also allows searching the entire base of importedterminologies.

123


Query authoring: Snow Owl includes an advanced editor for Extended SNOMED CTCompositional Grammar (ESCG) language expressions. The editor provides contextualsupport for query authoring by prompting based on the ESCG language specification.Other features include: color highlighting, as-you-type validation; quick-fix suggestionsfor common errors, user-definable templates for common expression patterns, versioncontrol system integration, and more.

Query evaluation: Snow Owls query evaluation engine uses information from thesemantic cache, the classifier, and indexes to improve performance.

Value Sets: Value sets can be created either manually (extensionally) or automatically(intensionally) with the use of an ESCG query. Extensional queries can be importedusing the SNOMED CT Simple Type Reference Set. Intensional queries are compatiblewith the SNOMED CT Query Specification Type Reference Set and can be optionallyconverted to manual reference sets if desired.

Mapping: Mapping between two terminologies is possible using Snow Owls mappingfeature, which is compatible with the SNOMED CT Simple Map Type Reference Setformat.

Reporting: Snow Owl includes a reporting and charting engine, which uses Snow Owlssemantic cache to accelerate report execution. A variety of graphs and charts can becreated and exported in several formats including HTML, PDF, and Microsoft Officeformats.

Validation: An extensible validation framework allows additional validation rules tobe plugged in and configured. Individual validation rules can be configured to executein batch or single validation mode. Validation results are shown in concept editors andnavigators in addition to the validation results view.

Description Logic Classification: ELK, SnowOwls default reasoner, performs descrip-tion logic classification in parallel on modern multi-core computers. This allows the fullinternational SNOMED CT plus the Australian extensions (830,926 relationships) to beclassified and checked for equivalencies in about 10 s on a modern desktop computer.

Task Management: SnowOwl integrates with existing issue tracking systems to display,search, assign, and schedule issues and tasks.

Collaborative Editing: Task workflow support enables collaborative authoring. Indi-vidual tasks can be activated and associated terminological changes can be grouped intochange sets associated with the tasks.When the workflow step is complete, the change setis provided as a context to allow the next participant in the process to easily continue theprocess. Upon final approval, the change set is committed to the terminology repository.

SNOMED CT Support: Working with SNOMED CT is simplified by providing a cus-tomized concept editor, navigator, advanced search, and query interfaces. AnatomicalTherapeutic Chemical (ATC) code Support: Snow Owl supports mapping to ATC codeswith a customized navigator and search interface.

Anatomical Therapeutic Chemical (ATC) code Support: Snow Owl supports mapping toATC codes with a customized navigator and search interface.

Extensibility: The Snow Owl framework can be extended by plugging in new terminol-ogies, new editors, new navigation views, new classifiers, and more.

4.3 OWL diff (2010)

This tool is designed for comparing OWL ontologies. It is of type view and TabWidget. Itaims to help managing and updating ontologies, which are often modified by several sides,

123


andmerging of concurrent updates is necessary. Usually, as for case of textual source codes, aversioning system is used for similar purpose: whenmultiple people update the same file, oneof them performs a merge, using a diff utility, which allows users to view the changes inthe file, select the changes which are to be included in the result, merge the files, and committhe resulting file. OWLDiff serves the same purpose for ontologies, as diff does for textualfiles. It takes two ontologies as arguments; known as the original and the update. Then it usesthe Pellet reasoner to check, if the two ontologies are semantically equivalent. If not, it showsthe differences graphically in two trees, one for each ontology. User can select differing itemsin either tree, which is to be updated in the resulting merged ontology.

The following features are the highlights of OWL Diff:

Left paneOverview of axioms of the old ontology. Right paneOverview of axioms of the new ontology. Main menuto adjust type of view, switch off reasoner and select different ontologies Select all/Deselect Allto select/deselect all axioms, which are different Mergeperforms merge on loaded ontologies with respect to selected axioms to be

deleted/removed CEXShows differences based on results of the CEX algorithm. Axiom propertiesshows the role of each axioms

Inferredthis axioms is not presented in the ontology, but can be inferred. The axioms,which were used for inferring are shown.

Axiom is in both ontologies Axiom has no connection to other ontology

4.4 OWL2Query (2011)

This is a conjunction query cum meta query engine and visualization plug-in. It facilitatesthe creation of queries using SPARQL or intuitive graph based syntax and evaluates themusing any OWL API complaint reasoner. This tool belongs to the category TabWidget andapplication. This tool incorporates several new features such as toolbar, prefix editor, vari-able editor, layout editor, query graph, SPARQL query view, SPARQL-DL preview, resultpanel, edge editor, property editor and Abox, Tbox & Rbox node editors. These help thedevelopment of effectiveness of users cognitive support.

4.5 OntoGraf (2010)

This tool is designed for Protg-OWL application. It offers support for interactively navigat-ing the relationships of created OWLontologies. It supports various layouts for automaticallyorganizing the structure of designed ontology. It is compatible with Protg-OWL 4.1 and4.2. It is available in various versions like1.0.1, 0.0.5, 0.0.4, 0.0.3, 0.0.2 and 0.0.1. The 1.0.1version fixed a weird generics problem with Java6. On the other hand the 0.0.5 version addedsupport to export the current visible graph into a DOT language format file. The addition ofnew view for visualization OWL imports and adds support for pinning the tool tip display toshow multiple tool tips at once and this is offered by version 0.0.4. By the mean time 0.0.3version added new tool tips exporting graphs as images and saving/opening graphs. Thistool incorporates additional features such as focus on home, grid alphabet, radial, spring,tree-vertical & horizontal directed, zoom-in, zoom-out, no-zoom, node-type, arc type andsearch (contains, start with, end with, exact match, reg exp).

123


4.6 OWLPropViz (2008)

This tool enables graphical representation of OWL object properties. Here users can selectthe property or properties to be visualized in the graph. The user can choose the class-axi-oms is-a, disjoint-with and equivalent-to. It is compatible with Protg-OWL 4.0. Itsdependencies are OWLViz and GraphViz. It is of type TabWidget.

4.7 PromptViz (2005)

This tool creates visual representations of the differences between two versions of an ontol-ogy. This is of type TabWidget. This is designed for the applications of Protg-Framesand Protg-OWL. It is compatible with Protg-Frames 3.3.1 and Protg-OWL 3.3.1. Thedependencies of this tool is PROMPT plug-in. PROMPTViz builds on the diff functionalityof the PROMPT tab for Protg by leveraging the difference table generated by PROMPTto create a treemap display showing a merge of the two versions of the ontology. The tree-map layout generates a space-constrained visualization of hierarchical structures and is veryeffective in showing attributes of leaf nodes using size and color coding. The treemap layout,combined with a zoomable user interface presents an initial overview that allows users todrill in and discover more details about the changes. The space occupied by a node in atreemap can be based on any numerical value. For example, the nodes could be sized bythe number of descendents they have or by the number of changes. An additional benefit ofthe treemap layout is that one can paint the nodes to provide additional information about thechanges to the node or its descendents. PROMPTViz does this in two ways. One is to paintthe node a solid colour representing the type of change that has happened to the class; theother way is to paint nodes with a histogram where each bar represents the percentage ofdescendents classified as each change type. PROMPTViz also has the ability to draw arcs toshow relationships between concepts and to depict the movement of concepts within the is-ahierarchy.

4.8 OWLViz (2010)

This tool enables class hierarchies in an OWL ontology to be viewed and incrementallynavigated. It also allows comparison of the asserted class hierarchy and the inferred classhierarchy. Its dependency isGraphViz.OWLViz also integrateswith the Protg-OWLeditor,where it uses the samecolour scheme so that primitive anddefined classes canbedistinguishedand computed changes to the class hierarchy may be clearly seen. It highlights inconsistentconcepts in red colour. It has the facility to save both the asserted and inferred views ofthe class hierarchy to various concrete graphics formats including PNG, JPEG and SVG. Itprovides asserted mode and inferred mode options. The cognitive support is enhanced by theuse of following options available with OWLViz: show class, show children, show parents,show all classes, hide class, hide children, hide classes past radius and hide all classes.

4.9 OWL2UML (2009)

This tool especially generates UML diagram that represents active ontology using OMGs(Ontology Management Group) Ontology Definition Meta model. This tool is compatible

123


with Protg OWL 4.0. It is of type view. The new versions added the features like fixationof some check box state bugs, change of graph layout file structure and fixation of someminor bugs. In addition, they also support inclusion of OWL: oneofvisualization and OWL:Restrictionvisualization. The highlights of this tool are as follows: selected entities, show:object property, command, associations, restrictions and generalizations.

4.10 NavigOWL (2011)

It is a visualization tool which is specially designed to example the semantic nets. The toolis enriched with appealing graph layouts that can be applied over the semantic net in order tounderstand the structure of ontologies easily. It also facilitates the user to build mental map inmore clear and consistent view of ontology graph. This tool supports RDF and OWL ontol-ogies to visualize them. Protg-OWL 4.1 and JDK 6 are its dependencies. This tool loadsRDF/OWL ontology file and configures its graph by rendering nodes and edges, based uponrole relations defined in ontology taxonomy. Its rendering factors configure various nodetypes. It also supports complete scalable directed graphs. It can visualize whole role relationshierarchy defined in ontology. It will show tool-tip exhibiting complete role-relation hierar-chy whenever one activities mouse-over event. It offers zoomable user interface. It providesgraph overview to show holistic view of large scale graphs to traverse through whole graph.In addition, this tool also facilitates users to apply various drawing layouts techniques toproduce appealing symmetric results of whole graphs. Node search and node label visibilityare the extended features supported by this tool. The Power Layout technique used in thistool, produces appealing drawing based upon node-degree distribution to understand nodesimportance.

4.11 Matrix (2010)

This tool provides several spread sheetstyle views of an ontology, including existentialfillers, individual relationships and an object properties view. It belongs to the type TabWid-get and view. It is designed for the application of Protg-OWL. This tool has the followinghighlights:

Class matrix (with inferred hierarchy) To add annotation column to matrix To add object property column to matrix To add data property column to matrix To remove column from matirx

Property matrix Object property matrix

add annotation column tomatrix, select property features column, remove columnfrom matirx

Data property matrix add annotation column tomatrix, select property features column, remove column

from matirx

Individuals matrix Individual matrix

123


add annotation column to matrix, add object property column to matrix add dataproperty column to matrix, remove column from matirx

Types matrix add annotation column to matrix, remove column from matirx

The following Table 1 summarizes features supported by various ontology visualizationtools.

4.12 DL Query (2008)

It offers the facility of quick test definition of classes to see that they subsume the appropriatesubclasses or to test for class membership of arbitrary descriptions without having to createnamed class. This tool belongs to the type TabWidget. It is designed for the application ofProtg-OWL. This tool adds effectiveness to the users cognitive support by introducingthe following features: Query box, execute option, object properties, class hierarchy win-dow, data properties, checkboxes for super class, ancestor class, equivalent class, subclass,descendent class and individuals and Query result frame.

4.13 Altova SemanticWorks

Altova SemanticWorks 2012 is the ground-breaking visual RDF and OWL editor for theSemantic Web. Graphically designed RDF instance documents, RDFS vocabularies, andOWL ontologies, then output them in either RDF/XML or N-Triples formats. Semantic-Works make the job easy with tabs for instances, properties, classes, etc., context-sensitiveentry helpers, automatic format checking, andmore. RDFS vocabularies define the allowableproperties (predicates) for RDF instances within a particular domain. RDFS also allow todefine classes to further classify the relationships between resources. Semantic Works dis-plays the instances, properties, and classes in an RDFS vocabulary on separate tabs, allowingone to view and edit these different items with ease.

The Instance tab lists all resources in the document, and the Properties tab lists all theproperties. When a property is selected in the main pane, the domain of that property is dis-played in another window. The Class tab lists all the classes available in the vocabulary witha separate window that lists the instances and properties of the selected class as they appearthroughout the RDFS. One can view and edit the details of any item listed by clicking itsexpand button. SemanticWorks display resources graphically according to their associationswith other resources. The SemanticWorks display is highly configurable. One can adjust thewidth of the items in the graph, display it with a vertical or horizontal orientation, adjustthe distances between parent and child nodes, and even change the font styles and colorsused. To help immediately visualize class relationships, RDFS classes are enclosed in yellowboxes in the graphical display. Holding the mouse over any item or icon display reveals itsmeaning or corresponding URI (Uniform Resource Identifier). The same entry helpers andcontext-sensitive choices described in the RDF editing section above are available for RDFSediting, and syntax checking based on the RDFS specification ensure that ones documentis valid. To creating a visual RDFS design, SemanticWorks 2012 is auto-generating thecorresponding RDF/XML or N-Triples code behind the scenes, and one can view and edit itat any time by clicking the Text tab.

123

Recent ontology visualization tools 607Ta

ble

1Su

mmaryof

features

supp

ortedby

variou

son

tology

visualizationtools

S.No

Tool

Type

Forapplication

Version

Com

patib

ility

Dependencies

Cog

nitiv

esupp

ort

features

1SO

VA(201

1)TabW

idget,view

Protg-OWL4.0

SOVA0.7.6

Protg-OWL4.1

Hermireasoner

Fullon

tology

visualization

Protg-OWL4.1

Protg-OWL4.2

Protg-OWL4.2

SOVA0.7.5

Protg-OWL4.0.2

SOVA0.6.0

Protg-OWL4.0

SOVA0.5.0

Protg-OWL4.0.2

2Sn

owOWL(201

1)App

lication

Protg-OWL4.1

Snow

OWL1.0.0

Protg-OWL4.1

Multi-platform

brow

ser

3OWLDiff(2010)

view

,TabWidget

Protg-OWL

OWLDiff0.1.4

Protg-OWL4.0.

OWLAPI

3.0.0

Ontologies

comparison

optio

nAny

OWL2reasoner

4OntoG

raf(2010)

TabW

idget,view

Protg-OWL

OntoGraf1.0.1

Protg-OWL4.2

Interactivenavigatio

n

Protg-OWL4.1

OntoGraf0.0.5

Protg-OWL4.1

OntoGraf0.0.4

Protg-OWL4.1

OntoGraf0.0.3

Protg-OWL4.1

OntoGraf0.0.2

Protg-OWL4.1

OntoGraf0.0.1

Protg-OWL4.1

5OWLProp

Viz(200

8)TabW

idget

ProtgOWL

OWLProp

Viz1.0

Protg-OWL4.0

OWLVizGraph

Viz

Objectp

roperties

Graphicalpresentatio

n

6Prom

ptViz(2005)

TabW

idget

ProtgFram

esProm

ptViz1.0.3

Protege-Fram

es3.3.1.

PROMPT

plug-in

Visualizationof

differencesof

twoversions

ProtgOWL

Protg-OWL3.3.1

7OWL2U

ML(2009)

view

Protg-OWL4.0

OWL2U

ML0.0.3

Protg-OWL4.0

Activeontology

representaion

OWL2U

ML0.0.2

Protg-OWL4.0

OWL2U

ML0.0.1

Protg-OWL4.0

123


Tabl

e1

continue

S.No

Tool

Type

Forapplication

Version

Com

patib

ility

Dependencies

Cog

nitiv

esupp

ort

features

8NavigOWL(201

1)TabW

idget,view

Protg-OWL

NavigOWL1.1.0

Protg-OWL4.1

Protg-OWL4.1

App

ealin

ggraphlayo

uts

Protg-OWL4.1

JDK6

9OWL2Q

uery

(201

1)TabW

idget,application

Protg-OWL

OWL2Q

uery

1.0.0

Protg-OWL4.1

OWLAPI

3.2.2

Graph

basedqu

erysyntax

Java6

10Matrix(2010)

TabW

idget,view

Protg-OWL

Matrix1.1.2

Protg-OWL4.2

Sp

read

sheet

styleview

s

Protg-OWL4.1

Matirx1.0.15

Protg-OWL4.0.2

11DLQuery

(2008)

TabW

idget

Protg-OWL

DLQuery

1.1.0

Protg-OWL4.2

Classes

testdefin

itons.

Protg-OWL4.1

DLQuery

1.0

Protg-OWL4.0

123


Fig. 5 FlexViz demo

4.14 FlexViz (2010)

FlexViz is a graph based visualization tool written is Adobe Flex. It supports single ontologybrowsing. Nodes of graph are mapped as concepts and relationships (edges) between nodes(e.g. is_a, depends_on) are represented as arcs. It is designed to provide a light-score,interactive, and visually accessible ontologies on Web (Sean et al. 2009). Figure 5 shows ademo of FlexViz, web-based visualization primarily render static images or text represen-tations. However, FlexViz greatly enhances user tasks and promotes new technologies bymaking it very interactive and light-weight.

FlexViz has many interesting features and can be summarized as follows:

Filtering based on nodes (concepts) and edges (relationships). Searching. Results may be viewed in various graphical layout. Customization of node and edge (arc) labels is possible. Customizing node and arc tool tips. Zooming. Forcibly staying node in the screen (can cause nodes to overlap). Back and forward button to navigate history. Nodes can be expanded or collapsed to show or hide children.

123


Colors of nodes, edges are customizable. Visualization can be displayed as a widget on a web page with a fixed ontology. Graph can be exported as a image data or a xml file.

4.15 OntVis (2006)

The class and property trees are crucial for making diagrams in OntVis. The toolbar helpsto change the size of the graph, and also lets switch between the pointer and the eraser. Toview relations and annotations for resources, the users can use the three-bar menu symbolthat appears when the mouse hovers over the resource. The user can do the following actionswith classes.

Show related classes Places subclasses, disjoint classes, and other related classes on the graph.

In some cases, the relations that appear will make use of resources already on thegraph.

Show annotations Expands the resources q2wsdef;klopcontainer to show its OWL annotations.

Show properties Expands the resources container to show related properties. The properties appear inside the resource box, but theymay be dragged onto the graph.

Note that properties, when dragged onto the graph, will appear as edges rather thanas nodes. If one can want to view a property as a resource, one must drag it out fromthe tree view.

Related properties are properties whose domain is declared to be the class in question,as well as properties with property restrictions declared on the class.

The following options are available with Properties:

Show related classes Places subproperties, inverse properties, and other related properties on the graph. In some cases, the relations that appear will make use of resources already on the

graph.

Show annotations Expands the resources container to show its OWL annotations. Show domain and range

Places the classes declared to be the domain and range of the property in question onthe graph.

On the other hand, the following options are available with instances

Show related instances Places related instances on the graph.

Show class assertions Expands the resource box to show which classes the instances is a member of.

123


Class assertions may be dragged on to the graph. Show property assertions Shows properties asserted on this instance and their values.

Property assertions may be dragged on to the graph.When dealing with complex classes or property restrictions, OntVis will display relevantinformation by adding a graphic to the edge. In some cases, letting the pointer hovers overthe edge will reveal a pop-up box with more information. Once the graph looking the wayone can want, one can choose Export under the File menu. One can choose PNG or PDFas the format.

4.16 Ontopia (2011)

Ontopoly is built as a client/server application. As a client, to use web browser, while theserver is a web server bundled with the distribution. The server-side application is builtusing the Navigator Framework and Web Editor Framework, which are parts of the Ont-opia Knowledge Suite. Ontopoly is accessible within Ontopia, an application that provideseasy access to Ontopoly, Omnigator, and Vizigator. General concepts:Ontopolys primarypurpose is to enable the manual creation and maintenance of topic maps that may be basedon a variety of ontologies. In order to be able to provide the most intuitive possible userinterface for such a generic application, Ontopoly is ontology-driven. What this means isthat the forms-based interface for creating and maintaining a topic map is generated auto-matically from the underlying ontology and the rules that are defined for it. For this reason,to create an ontology before one can edit a topic map instance in Ontopoly. Often the tasksof creating the ontology and editing the topic map instance (that is, populating the ontol-ogy) will be carried out by different people. Ontopoly is therefore divided into two mainparts: the ontology editor and the instance editor. The application also has an administrationinterface, i.e., a page for adding metadata to the topic map (description, creator, version,etc.), validating it, etc.; and an interface for exporting to various interchange syntaxes. TheTopic Map Index Page allows the user to open an existing topic map, create a new one, orimport one from outside Ontopoly. Once selected one of these actions, then it will take to theapplication pages. Ontopolys two primary functions provide a comprehensive Topic Mapsediting environment.

4.16.1 Ontopoly topic maps

A topicmap created in Ontopoly (anOntopoly topic map) will differ from one created outsideof it (a non-Ontopoly topic map). An Ontopoly topic map carries along with it topics thatlet it interact with the Ontopoly application and topics that define the schema (the systemtopics). One can always export a topic map from Ontopoly to remove the system topics, butthey are needed for the topic map to be understood within the application similarly, pre-existing non-Ontopoly topic maps will need these system topics to be added. There are fivekinds of types in Topic Maps: topic types, occurrence types, association types, role types,and name types. The properties that are common to all typing topics and thus occur on allConfiguration Pages. This applies to Name, Subject identifier , Description, Read-only,and Hidden.

123


Fig. 6 Notes types with the icons

4.17 Collaborative Protg (2009)

Collaborative Protege is an extension of the existing Protege system that supports collabo-rative ontology editing. In addition to the common ontology editing operations, it enablesannotation of both ontology components and ontology changes. It supports the searchingand filtering of user annotations, also known as notes, based on different criteria. One canimplement two types of voting mechanisms that can be used for voting of change proposals.Multiple users may edit the same ontology at the same time. In multi-user mode, all changesmade by one user are seen immediately by other users. There are twoworkingmodes availablefor Collaborative Protege. Both modes support multiple users working on an ontology:

The multi-user modeallows multiple clients to edit simultaneously the same ontologyhosted on a Protege server. All changes made by one client are immediately visible by otherclients. This mode is also referred to as client-server mode, or concurrent mode and requiresa client-server setup. This mode is based on the implementation of the multi-user Protege andis the preferred mode in which Collaborative Protege should be run.

The standalone modeallows multiple users to access the same ontology in succession.The ontology can be stored on a shared network drive and all clients will access the sameproject files. However, simultaneous access is not possible. This mode is also referred to asthe consecutive mode.

Themain feature of Collaborative Protege is the ability to create notes attached to differentthings. The notesmechanism of Collaborative Protege allows a user to create his own remarksabout a certain part of the ontology. This feature can also be used to discuss the ontology withother users either in stand-alone or multi-user mode. The notes are also called annotation.The notes types together with the icons used to represent them in the user interface are shownin Fig. 6.

The main functionalities of Collaborative Protg are: Annotation of classes, propertiesand instances with different types of notes (e.g., Comment, Advice, Example, etc.)

Annotation of changes (for example, class creation, renaming, etc.) with different typesof notes

123


Notes on the ontology itself Proposals and voting Searching and filtering the notes using different criteria Live discussion (chat) with other users connected to the same server The main panel of the graphical user interface has been divided in two main panels : The classical Protege display (classes, properties and instances view) on the left side, and The collaboration panel on the right side.The classical Protege display shows the details of the different ontology components, such asclasses, properties (slots) and instances. The collaboration panel adds functionality to supportthe collaborative development of ontologies.

The collaboration panel is made up of two types of tabs.

The Entity notes and Changes tabs show information related to the selected item in theclass tree (or property tree, etc.). For example, if the user select in the class tree theclass Person, the user will see in these tabs the notes related to the class Person. Inthe InstancesTab of the main Protege display, the collaboration panel will show the notesattached to instances.

The Ontology notes, All notes, Search and Chat tabs are generic tabs that are not attachedto a particular entity in the ontology, so they are global in nature.

The tabs in the collaboration panel are made up of a top component, which shows a notes tree.The notes tree corresponds to discussion threads in Internet forums. The bottom componentshows the details of the selected note in the notes tree.

It is possible to search the notes by using the search component at the bottom of the anno-tations tree panel. One can use the * wildcard character for searching process. The searchingis done in the text of the notes.

Figure 7 shows an example of notes attached to a specific ontology component in theEntityNotes. By ontology component (classes, properties (or slots) and instances). Another featureis the filtering of notes and changes, which may be done by using the Filter component fromthe top of the annotations tree. One can choose between different types of filters that willbe applied to the annotations tree: filtering by author, by annotation text, by annotation typeor by creation date. The class tree in the main Protege display shows a little call-out iconnext to the class name if the class has any annotations or change annotations attached to it.In the Slots Tab and Instance Tab in the main Protege display user will also see the call-outicon if a property (slot) or an instance have annotations attached to them.

A user may add notes attached to a specific ontology component in the Entity Notes. Byontology component, to mean classes, properties (or slots) and instances. The image showsan example of note attached to the DomainConcept class. The number of notes attacheddirectly to DomainConcept is shown next to the Notes icon.

4.18 Wandora (2007)

Wandora is a general purpose information extraction, aggregation, management, and publish-ing application based on Topic Maps and Java. Wandora has graphical user interface, layeredpresentation of knowledge, several data storage options, huge collection of data extraction,import and export options, embedded server, and open plug-in architecture. Wandora is aFOSS application with GNU GPL license. Wandora is well suited for constructing informa-tion mashups and is capable of extracting and converting a wide range of open data feedsto Topic Map formats. Beyond Topic Maps conversion this feature allows Wandora user to

123


Fig. 7 Collaborative pizza.owl Protg 3.4.2

aggregate multidimensional information mashups where information from Flickr interleaveswith information from Geo Names and YouTube, for example. Wandoras embedded HTTPserver enables easy mashup publication.

4.18.1 GraphXML export

Wandora can export Topic Maps in GraphXML format. Exported graph can be visualizedwithHypergraph application for example. Below is an example ofHypergraph applet viewingWandoras base ontology exported as GraphXML (Fig. 8).

Export details

Each topic is exported as a graph node. Each binary association is converted to an edge between player nodes. Edges label is

association type. If association has more than two players, an extra intermediate node is created and each

player is linked to the intermediate node with an edge. If association has only one (or none) player, it is not exported. Each occurrence is exported as a node and an edge between occurrence text and topic

node. Subjects, scopes, association roles, and variant names are not exported. Exported nodes and edges dont include coordinates.

123


Fig. 8 Wandoras base ontology exported as GraphXML

4.19 TopBraid composer (2008)

It is an enterprise-class modeling environment for developing Semantic web ontologies andbuilding semantic applications. It is fully complaint with W3C standards. It offers compre-hensive support for developing, managing and testing configurations of knowledge modelsand their tool is an industrialstrength RDF editor as well as the SPARQL tool. It also incor-porates a flexible and extensible frame work with a published API for developing Semantic.Semantic client/server of browser-based solutions that can integrate disparate applicationsand data sources. It is used to develop ontology models, configure data source integration aswell as to customize dynamic forms and reports.

4.20 Knoodl (2009)

It facilitates communityoriented development of OWL based ontologies and RDF knowl-edge bases. It also offers a JAVA service-based interface or a SPARQL-based interface. It isa product of Revely tix. The main capabilities include.

Cloud-based application Ontology editing Ontology import / export Collaboration and querable threaded discussing Role-based security Choice of scalable RDF stones Ontology guided search SPARQL endpoints HTTP service for graph updates SPARQL query wizard User designed views of query results.

123


4.21 NeOn toolkit (2011)

It is a open source multi-platform ontology engineering environment. It provides compre-hensive support for the ontology engineering life-cycle. The toolkit is based on the Eclipseplatformand it provides an extensive set of plug-ins covering avariety of ontology engineeringactivities, including Annotation and Documentation, Development, Human-Ontology inter-action, Knowledge Acquisition, Management, Modularization and customization, ontologyDynamics, Ontology evaluation, ontology matching, Reasoning and interface and reuse.

4.22 pOWL (2004)

pOWLSemantic Web Development Platform (http://powl.sf.net/) has the aim to deliver aPHP and Web-based ontology editing and management solution to the open source commu-nity. The application is aWeb-based one and supports viewing, editing of RDFS/OWL ontol-ogies of arbitrary size. pOWL currently offers an RDQL query builder as well as a full-textsearch for literals and resources. pOWL offers support for extensions, butunfortunately still lacking exhaustive documentation on this. All functionality is accessibleby a solid application programming interface (API) and the access could be authenticated.Models are stored in database tables (e.g., MySQL, PostgreSQL, Oracle, etc.) and only thoseparts of the model are loaded into main memory which are actually neededthis assures afast response. The ontology can be imported from and exported to different RDF syntaxes(XML, N3, Ntriples). Also, the results can be available as HTML pages. pOWL is freelyavailable under the terms of GNU General Public License (GPL).

4.23 SWOOP (2006)

SWOOP (SemanticWeb Ontology Overview and Perusal) (http://www.mindswap.org/2004/SWOOP/) is a simple, scalable, hyper media inspired OWL ontology browser and editorwritten in Java. Other familiar web-browser look and feel features include an address barto load ontological entities, history buttons and bookmarks. SWOOP has been designed in-keeping with the W3C OWL recommendations and has reasoning support (Pellet, an OWLinference engine). Another facility is the multiple ontology environments, whereby entitiesand relationships across various ontologies can be seamlessly compared, edited and merged.All ontology editing in SWOOP is done inline with the HTML rendered, using different colorcodes and font styles to emphasize ontology changes, e.g. diverse representations for added,deleted or inferred axioms. Undo/redo options are provided with an ontology change log anda rollback option. SWOOP could import ontologies fromOWL, XML, RDF and text formats.These formats could be used to save the edited ontologies. The overall tool architecture isbased on MVC (Model-View-Controller) design pattern.

4.24 GrOWL (2007)

It is based around the underlying DL semantics of OWL ontologies (Sergey et al. 2007).It has advanced browsing and navigation tools. GrOWL is implemented as a Java applet asa Protg plug-in and as a stand-alone Java application. Its editor uses Jena API, and the

123

http://powl.sf.net/http://www.mindswap.org/2004/SWOOP/http://www.mindswap.org/2004/SWOOP/


GrOWL applet uses a trimmed version of the wonder web OWLAPI. It uses the followingtechniques:

4.24.1 Automatic and natural layout

It uses animated force directed layout, which performs the layout interactively in combina-tion with animation. GrOWL also supports a manual layout to allow the user to repositionindividual nodes as they wish

4.24.2 TBox, ABox, RBox and class hierarchy views

There are several meaningful ways to filter out a part of the entire graph, including viewingonly the ABox, TBox or class hierarchy. Such filtering may not require selection of a focalnode.

4.24.3 Locality restricted view

An attempt at drawing even a simple semantic network on a two-dimensional surface wouldlikely end up with a non-planar graphmeaning that the relationships that link conceptstogether would be drawn crossing and overlapping each othera very messy and potentiallyunusable view of the underlying structure. It is possible to provide a mechanism for restrict-ing the set of visible nodes and edges to only those that lie in a topological neighborhoodof a selected node. When the selected node is changed the set of visible nodes and edges isreadjusted accordingly. A node may be set as selected by clicking on it or by obtaining it asa result of one or another search and navigation method.

4.24.4 Tree based navigation

The list of classes and instances occurring in the ontologies may be given in a class treefamiliar to the users of Protg. Selecting any object from this tree sets the respective node asselected. This action readjusts the visible locality of the network, centering it on the selectednode.

4.24.5 Filtering

GrOWLprovides amechanism for restricting the view to only class definition, the subclasses,the super classes, or all instances associated with a selected node. This filtering mechanismis complementary to the topological neighborhood mechanism. Filters make the presentationmeaningful by allowing the user to focus precisely on what is of interest. The filter thatoutputs the class definition for the selected class node has proven to be specially useful.

4.24.6 Prefix search

Users can search for the nodes (classes and properties) using incremental matching. Whileuser types the name of the node the information about the set of nodes with the given prefixis displayed. Selecting one of the nodes brings it into focus and the visible locality of thenetwork is readjusted using the current filter.

123


5 Evaluation descriptions

An experiment was conducted in order to evaluate users satisfaction on cognitive supportimplemented by three different ontology visualization tools. As a part of our real time project,we created ontologies for HIV-AIDS patients information. The experiment described in thiswork was designed in order to provide useful insight concerning three research areas, whichare:

The evaluation of 3 ontology visualization tools- DL Query, OntoGraf and OWL2Query. The strategies and techniques employed by the users while researching HIV information. The evaluation of the HIV-AIDS ontology created by our research project.

This paper limits the discussion to the results concerning the advantages and disadvantagesof the three ontology visualization tools. The focus of this experiment was not overall ontol-ogy management and editing, but rather information retrieval and assessing the stability ofeach method for end user applications where ontologies are used as browsing assisters. Thissection overviews the performed evaluation, containing brief descriptions of the evaluationuser group, the ontology used, the query types used for information retrieval through theontology, the description of the evaluation sequence and the results.

5.1 Test user group

In order to examine the effectiveness of the evaluated ontology visualization, a user groupwith both computer skills and basic domain knowledge was choosen. The choice of ontologywas such that all the users could have at least some familiarity with computers and domain.This fact ensured that there would not be significant differences in the performance of theusers due to complete lack of knowledge of the domain. Then the ontology created for theHIV-AIDS domain was choosen.Most of the users that participated to the experiment wereresearch scholars of Computer Science and Information Science departments of Sri PushpamCollege of Bharathidasan University, India. All these users have some knowledge both incomputers and HIV domain as they are instructed to be familiar for testing. But they varyin degree of skills both in computer and domain knowledge respectively.The user group wasdivided into two commensurate groups of eight users one that got a short introduction in howto use these testing tools, and one without any prior knowledge about ontology editing.

5.2 About the ontology used

The ontology used in this experiment is an effort to describe the domain of the HIV-AIDSTamilnadu. It presents the current AIDS patients information under the treatment and alsocontains relevant information about the symptoms and causes of HIV. It contains 15 classes.It is populated with 2,573 instances. The maximum depth of the is-a taxonomy tree is 13classes. Multiple inheritance has been employed for about 2 classes and no other classes arehaving more than one parent.

5.3 Testing method set-up

Before the start of the evaluation we had to perform some preliminary tests in order to decideupon the visualization method set-up to be used in the experiment.

123


Bearing inmind that wewere investigating themost suitable visualization not for ontologydevelopers but for users that will use the ontology as an information retrieval aid, we hadto keep the visualization method controls as simple as possible. Furthermore, for the sizeof the experiment ontology, some visualization set-ups were really cluttered and not at alluseful for information retrieval. In the case of DLQuery, Query box, execute option, objectproperties, class hierarchy window, data properties, checkboxes for super class, ancestorclass, equivalent class, subclass, descendent class and individuals and Query result framewere introduced to the users. In the case of OntoGraf, focus on home, grid alphabet, radial,spring, tree-vertical & horizontal directed, zoom-in, zoom-out, no-zoom, node-type, arc typeand search (contains, start with, end with, exact match, reg exp) options were introduced tothe users. Finally, for the case of OWL2Query, we introduced the following features to theusers: toolbar, prefix editor, variable editor, layout editor, query graph, SPARQL query view,SPARQL-DL preview, result panel, edge editor, property editor and Abox, Tbox & Rboxnode editors.

5.4 HIV-AIDS ontology information retrieval tasks

This experiment constructed five different queries to retrieve information from HIV-AIDSontology. The queries are then grouped into different types according to ontology related cri-teria, such as the number of different classes they entail, if they are relevant to the ontologyhierarchy or not, if they ask for the number of classes of instances with a common charac-teristics etc. The identified query types are presented in the following text, along with briefdescription examples.

1. The user is given the value of a slot of an instance, and is asked to find the value ofanother slot of the same instance. For example, What is the year of registration of thepatientid 101?In this case the user has to locate a specific instance and then extract a slot value to findthe answer to the query.

2. The user is given the value of a slot of an instance I1, and is asked to retrieve a slot valueof some instance I2, linked to I1 through a role relationship. For example, What is theyear of identification of the HIV +ve that a particular patient affected with?In this case the user should first locate I1, follow the role relationship to I2 and thenextract a slot value to find the answer to the query.

3. Query related to the class hierarchy, the taxonomy. In this case, a class is described tothe user and he/she is asked to retrieve its direct subclasses. For example, What are thesymptoms of the HIV +ve?In this example, the result is a set of class names, which should be organized hierarchi-cally.

4. Querying for the number of instances of a specific class. For example, What is thenumber of the drug manufacturer?In this case the result is a number, so the user has to locate the specific instances andcount them or view their number if this feature is provided by the interface.

5. Retrieve the number of instances with a specific common slot value. For example, Whatis the number of patients prescribed with a specific drug?In this case, as in query 4 of the previous section, the result is a number, so the userhas to locate the specific instances and count them (or view their number if this featureis provided by the interface). However, this case is somewhat more complicated as notall the instances of an entity are requested, but a sub-set of them with a common slot

123


Fig. 9 TLX and time comparison for five different tasks in OntoGraf, DLQuery and OWL2Querlabel

value. The user groups were involved in retrieving the answers for those queries usingDL Query, OWL2Query and OntoGraf.

5.5 Performance analysis

For each task (query), we measured the NASA Task Load Index (TLX) (Sandra and Lowell1988) and the time that was needed to perform the task. Figure 9 shows the results of thecomparison of OWL2Query, DLQuery and OntoGraf ontology visualization tools based onthe visualization scores as perceived by the users. As we expected, we can see that the group

123


that was given a short introduction into the tools, performed better on average. It scored lowerTLX and shorter time. For task1, there was almost no difference observed between the twogroups with respect to OntoGraf. The Fig. 9 clearly shows OntoGraf let to a lower TLX onaverage in all five tasks compared to other two tools such as OWL2Query and DLQuery, thatis, users were less frustrated and mentally stressed. That is very likely also an outcome ofthe reduced time users had to spend for each task. The figure shows that users could solveeach task faster using OntoGraf. On average users spent approximately 59.6% less time withOntoGraf and had a 25.4% lower TLX than when using OWL2Query and 33.2% less timewith OntoGraf and had a 16.4% lower TLX than when using DLQuery. The comparison onlyshows that OntoGraf is better suited for the use case of HIV-AIDS ontology we described.DLQuery and OWL2Query are however much more feature enriched tools which make itdifficult to use than OntoGraf for our experiment. We evaluated only information retrievalprocess that was possible with these three tools.

6 Conclusion

A survey of advances of cognitive support in recent ontology visualization tools is presentedin this paper. The review includes both research and commercial category tools. Tools ofsimilar purpose are choosen to analyze their features. This work also included in its anal-ysis information about the import/export format, graph view, consistency check, version,dependencies, type, multi-user type, web support, library support and etc., The result of thissurvey and analysis provides comprehensive understanding of new features that enhancecognitive support. Finally, we presented some preliminary results from a comparative eval-uation of selected three visualization tools. The results are being further analyzed in orderto extract interesting patterns. Furthermore, the results of this evaluation are being analyzedwith respect to the other two aspects of the experiment, i.e. the evaluation of the ontologyitself and a test of the methods users employ for dealing with various information retrievaltypes. This work can be extended with other tools/ frameworks for the complete ontologymanagement operations.

Acknowledgments This work has been financed by University Grants Commission Major Research Pro-jectRef. No. 38-4/2009(SR) (India).

References

Altova Semantic works:http://www.altova.com/semanticworks.htmlBenjamin, B, Emmanuel, P, Ilaria, L, Gennady, L (2011) OntoTrix: a hybrid visualization for populated

ontologies. In: Proceedings of WWW2011. HyderabadBerners-Lee T, Handler J, Lassila O (2001) The semantic web. Scientific AmericanBoinski T, Bundnik L, Jaworska A, Mrozinski J, Mazurkiewicz K (2009) OCS-domain oriented ontology

creation system. Pol J Environ Stud 18(3B):3538Boinski, T, Jaworska, A, Kleczkowski, R, Kunowski, P (2010) Ontology Visualization. In: Proceedings of the

2nd international conference on information technology, ICITDLQuery: http://protegewiki.stanford.edu/wiki/DL_QueryEades P (1984) A heuristic for graph drawing. Congr Numer 42:149160FlexViz:http://www.thechiselgroup.com/flexvizFurnas, GW (1986) The FISHEYE view: a new look at structured files. In: Proceedings of the conference on

human factors in computing systems. ACM, pp 1623Graham M, Kennedy J, Benyon D (2000) Towards a methodology for developing visualizations. Int J Hum

Comput Stud 53(5):789807

123

http://www.altova.com/semanticworks.htmlhttp://protegewiki.stanford.edu/wiki/DL_Queryhttp://www.thechiselgroup.com/flexviz


Guarino, N, Giaretta, P (1995) Ontologies and knowledge bases: towards a terminological clarification.Towards very large knowledge bases: knowledge building and knowledge sharing. IOS, pp 2532

Herman I,Melancon G,Marshall MS (2000) Graph visualization and navigation in information visualisa-tion:a survey. IEEE Trans Vis Comput Graph 6(1):2443

Johnson B, Shneiderman B (1991) Treemaps: a space-filling approach to the visualisation of hierarchicalin-formation structures. In: Proceedings of the 2nd international IEEE visualization conference. SanDiego

KAON:http://kaon.semanticweb.org/Katifori A, Halatsis C, Lepouras G, Vassilakis C, Giannopoulou E (2007) Ontology visualization methodsa

survey. ACM Comput Surv 39(4):143Katifori A, Torou E, Halatsis C, Lepouras G, Vassilakis C (2006) A comparative study of four ontology

visualization techniques in Protg: experiment setup and preliminary results. http://ieeexplore.ieee.org/stamp/stamp.jsp?arnumber=01648294

Knoodl. http://knoodl.com/ui/home.htmlKobsa A (2004) User experiments with tree visualization systems. In: IEEE symposium on information visu-

alization (INFOVIS04), pp 916Lamping J, Rao R, Pirolli P (1995) A focus+context technique based on hyperbolic geometry for visual-izing

large hierarchies. ACM conference on human factors in omputing systems (CHI95). ACM Press, NewYork, pp 404408

Munzner T (1997) H3: laying out large directed graphs in 3D hyperbolic space. In: Proceedings of the IEEEsymposium on information visualisation. Phoenix

Musen MA, Fergerson RW, GrossoWE, Noy NF, GrubezyMY, Gennari JH (2000) Component-based supportfor building knowledge-acquisition systems. In: Proceedings of the intelligent information processing (IIP2000) conference international federation for processing (IFIP), world computer congress (WCC2000).Beijing, pp 1822

Nadia C, Lorenzo R (2009) User-friendly ontology editing and visualization tools: the OWLeasyViz approach.In: Proceedings of the 13th international conference information visualisation

NavigOWL:http://protegewiki.stanford.edu/wiki/NavigOWLNeOn toolkit:http://neon-toolkit.org/wiki/Main_PageNorth C, Shneiderman B (2000) Snap-together visualisation: can users construct and operate coordinated

visu-alisations?. Int J Hum Comput Stud 53:715739North SC (1995) Incremental layout in DynaDAG. In: Proceedings of the symposium on graph drawing GD

95. Springer, Berlin, pp 409418Noy NF, McGuiness DL (2001) Ontology development 101: a guide to creating your first ontology, stanford

knowledge systems laboratory technical report KSL-01-05 and stanford medical informatics technicalreport SMI-2001-0880

Noy NF, Sintek M, Decker S, Crubezy M, Fergerson RW, Musen MA (2001) Creating semantic web contentswith Protege-2000. IEEE Intell Syst 16:6071

OilEd. http://oiled.man.ac.uk/OntoGraf:http://protegewiki.stanford.edu/wiki/OntoGrafOntopia:http://www.ontopia.net/OntVis: http://www.revelytix.com/content/revelytix-releases-ontvis-owl-ontology-visualization-both-busine

ss-people-and-ontologistsOWL2Query: http://protegewiki.stanford.edu/wiki/OWL2QueryOWL2UML: http://protegewiki.stanford.edu/wiki/OWL2UMLOWLDiff: http://krizik.felk.cvut.cz/km/owldiff/OWLPropViz: http://protegewiki.stanford.edu/wiki/OWLPropVizOWLViz: http://www.co-ode.org/downloads/owlviz/pOWL: http://powl.sf.net/PromptViz: http://www.thechiselgroup.com/promptvizprotg: http://protege.stanford.edu/Risden K, Czerwinski MP, Munzner T, Cook D (2000) An initial examination of ease of use for 2D and 3D

information visualizations of web content. Int J Hum Comput Stud 53:695714Sandra GH, Lowell ES (1988) Development of NASA-TLX (Task Load Index): results of empirical and the-

oretical research. Hum Ment Workload 1:139183Sean MF, Chris C, Margaret-anne S (2009) FLEXVIZ: visualizing biomedical ontologies on the web. In:

International conference on biomedical ontology. Software Demonstration, BuffaloSergey K, Richard W, Ferdinando V (2007) GrOWL: a tool for visualization and editing of OWL ontologies.

Journal of Web Semantics. Science Direct. Elsevier, Amsterdam, pp 5457SWOOP (Semantic Web Ontology Overview and Perusal): http://www.mindswap.org/2004/SWOOP/

123

http://kaon.semanticweb.org/http://ieeexplore.ieee.org/stamp/stamp.jsp?arnumber=01648294http://ieeexplore.ieee.org/stamp/stamp.jsp?arnumber=01648294http://knoodl.com/ui/home.htmlhttp://protegewiki.stanford.edu/wiki/NavigOWLhttp://neon-toolkit.org/wiki/Main_Pagehttp://oiled.man.ac.uk/http://protegewiki.stanford.edu/wiki/OntoGrafhttp://www.ontopia.net/http://www.revelytix.com/content/revelytix-releases-ontvis-owl-ontology-visualization-both-business-people-and-ontologistshttp://www.revelytix.com/content/revelytix-releases-ontvis-owl-ontology-visualization-both-business-people-and-ontologistshttp://protegewiki.stanford.edu/wiki/OWL2Queryhttp://protegewiki.stanford.edu/wiki/OWL2UMLhttp://krizik.felk.cvut.cz/km/owldiff/http://protegewiki.stanford.edu/wiki/OWLPropVizhttp://www.co-ode.org/downloads/owlviz/http://powl.sf.net/http://www.thechiselgroup.com/promptvizhttp://protege.stanford.edu/http://www.mindswap.org/2004/SWOOP/


Wandora: http://www.wandora.org/wandora/wiki/index.php?title=Main_PageWills GJ (1997) NicheWorksinteractive visualisation of very large graphs. In: Proceedings of the graph

drawing 97. Springer, Rome. pp 403414

123

http://www.wandora.org/wandora/wiki/index.php?title=Main_Page

A study on development of cognitive support features in recent ontology visualization toolsAbstract1 Introduction2 About Protg3 Ontology visualization methods3.1 The indented list method3.2 Node link and tree3.3 Zoomable method3.4 Focus+context method

4 Advances in cognitive support:recent ontology visualization tools4.1 SOVA (2011)4.2 Snow OWL (2011)4.3 OWL diff (2010)4.4 OWL2Query (2011)4.5 OntoGraf (2010)4.6 OWLPropViz (2008)4.7 PromptViz (2005)4.8 OWLViz (2010)4.9 OWL2UML (2009)4.10 NavigOWL (2011)4.11 Matrix (2010)4.12 DL Query (2008)4.13 Altova SemanticWorks4.14 FlexViz (2010)4.15 OntVis (2006)4.16 Ontopia (2011)4.16.1 Ontopoly topic maps

4.17 Collaborative Protg (2009)4.18 Wandora (2007)4.18.1 GraphXML export

4.19 TopBraid composer (2008)4.20 Knoodl (2009)4.21 NeOn toolkit (2011)4.22 pOWL (2004)4.23 SWOOP (2006)4.24 GrOWL (2007)4.24.1 Automatic and natural layout4.24.2 TBox, ABox, RBox and class hierarchy views4.24.3 Locality restricted view4.24.4 Tree based navigation4.24.5 Filtering4.24.6 Prefix search

5 Evaluation descriptions5.1 Test user group5.2 About the ontology used5.3 Testing method set-up5.4 HIV-AIDS ontology information retrieval tasks5.5 Performance analysis

6 ConclusionAcknowledgmentsReferences

a study on development of cognitive support features in recent ontology visualization tools

Documents