Chapter 9Conclusions

This book addresses the key technologies for developing an intelligent geospatialknowledge system based on Web services. In particular, Semantic Web technol-ogies are introduced into the area of geospatial Web services. The ontology-basedapproach helps to improve the recall and precision of data and services discoveryprovided by the catalogue service. Semantics-enabled metadata tracking and sat-isfaction allows analysts to focus on the generation of a geospatial process modelrather than spending large amounts of time in data preparation. ‘‘DataType’’-driven service composition and path planning can help to automate a range ofknowledge discovery processes in a limited geospatial domain. Process planningfacilitates the construction of complex services and models for geocomputation.The process models and service chains can be archived, catalogued, and advertisedas new geospatial services in CSW and thus be discovered and used for futuregeospatial modeling.

Semantics for geospatial data, services, and service chains, are organized andregistered in the CSW by extending ebRIM elements. In particular, the IOPEsemantics are introduced for geospatial services and the registration of IOPEsemantics is proposed in the CSW-ebRIM profile for loosely-coupled, mixed-coupled, and tightly-coupled services and data. A ProcessModel class is added tothe ebRIM model. It is a key extension, addressing the requirements for geospatialinformation processing and knowledge discovery purpose in a data-rich distributedenvironment, as opposed to the initially core extension class Dataset in the CSW-ebRIM profile. Middleware to support semantics-enhanced discovery of geospatialdata, services/service chains, and process models has been developed. Suchmiddleware can be applied to support materialization of virtual data products. Theapproach demonstrates that semantics can be used to improve the data/services/service chains discovery capability of geospatial catalogue services. The processmodels, working as a kind of geospatial knowledge, can address the analysis issuesin the Cyberinfrastructure and support on-demand delivery of geospatial infor-mation and knowledge.

The book shows that standards-based geospatial Web services can be used todiscover, request, access, and obtain geospatial data and information in a

P. Yue, Semantic Web-based Intelligent Geospatial Web Services,SpringerBriefs in Computer Science, DOI: 10.1007/978-1-4614-6809-7_9,� The Author(s) 2013

103

distributed environment. A general procedure for automatic geospatial servicecomposition is proposed. This procedure includes three phases: process modeling,process model instantiation, and workflow execution. The approach in this bookcan address all types of service chaining in the OGC abstract service architecture:user-defined, workflow-managed and aggregate. The existing process modelsallow analysts to interactively construct new, complex Web-executable geospatialprocess models.

Future work includes the application of the approach in various use casesinvolving more types of geoprocessing services and generation of more complexmetadata. Such an application should provide primarily semantic descriptions ofall involved services. For example, over one hundred geoprocessing services havebeen developed in GeoPW (Yue et al. 2010). How these services determine theappropriateness of starting data or affect the input data should be describedsemantically, explicitly, and formally. And we believe that specifying suchmetadata requirements and metadata changes by an appropriate semanticdescription of geoprocessing services is important to allowing automatic servicechaining and semantic evaluation of service chains in the future. In addition, theinsertion of data reduction and transformation services for satisfying metadataconstraints is implemented as domain control procedures by providing functions inthe computer source code. These domain control procedures embody domainknowledge, which can be expressed using some knowledge representationmechanism to facilitate AI planning. It is possible to transform the problem ofsatisfying metadata constraints to an AI planning domain so that traditionalAI planners can be used with semantic geospatial services to generate a plan forservice chains.

The CSW-ebRIM profile has been adopted because it allows catalogues tohandle geospatial data, services, and other types of information resources such asapplications schemas, software components, and reference documents; also, it canhandle process models, as shown in this chapter. This capability to register dif-ferent information resources is demonstrated through the employment of thoseextensibility points in this paper. There are other catalogue services, for example,the NASA EOS Clearinghouse (ECHO), the U.S. Department of Energy (DOE)Earth System Grid (ESG) Simulation Data Catalogue, discovery frameworks likeUDDI, and other application profiles of the OGC catalogue specification like theISO metadata application profile. Some of them differ in the catalogue querylanguage and communication protocol, while others differ in the informationmodel. If multiple catalogue services must be used, they can be federated toprovide a comprehensive discovery of geospatial information. It is possible to addthe semantic middleware in the federation service as a future work.

104 9 Conclusions