Semantic Description and Matching of Services for Pervasive Environments

Abstract

With the evolution of the World Wide Web and the advancement of the electronic world, the diversity of available services is increasing rapidly. This raises new demands for the efficient discovery and location of heterogeneous services and resources in dynamically changing environments. The traditional approaches for service discovery such as UDDI, Salutation, SLP etc. characterise the services by using predefined service categories and fixed attribute value pairs and the matching techniques in these approaches are limited to syntactic comparisons based on attributes or interfaces. More recently with the popularity of Semantic Web technologies, there has been an increased interest in the application of reasoning mechanisms to support discovery and matching. These approaches provide important directions in overcoming the limitations present in the traditional approaches to service discovery. However, these still have limitations and have overlooked issues that need to be addressed; particularly these approaches do not have an effective ranking criterion to facilitate the ordering of the potential matches, according to their suitability to satisfy the request under concern. This thesis presents a semantic matching framework to facilitate effective discovery of device based services in pervasive environments. This offers a ranking mechanism that will order the available services in the order of their suitability and also considers priorities placed on individual requirements in a request during the matching process. The proposed approach has been implemented in a pervasive scenario for matching device-based services. The Device Ontology which has been developed as part of this research, has been used to describe the devices and their services. The retrieval effectiveness of this semantic matching approach has been formally investigated through the use of human participant studies and the experimental results have indicated that the results correlate well with human perception. The performance of the solution has also been evaluated, to explore the effects of employing reasoning mechanisms on the efficiency of the matching process. Specifically the scalability of the solution has been investigated with respect to the request size and the number of advertisements involved in matching.