An Algebra of Ontology Properties for Service Discovery and Composition in Semantic Web

Abstract

We address in this chapter the problem of the autom ated discovery and composition of Web Services. Now, Service-oriented computing is emergi ng as a new and promising paradigm. However, selection and composition of Services to a chieve an expected goal remain purely manual and time consuming tasks. Basing our approac h n domain concept definitions thanks to an Ontology, we develop here an algebraic approach that enables to express formal definitions of Web Service semantics as well as user information n eeds. Both are captured by the means of algebraic expressions of ontology properties. We pr esent an algorithm that generates efficient orchestration plans, with characteristics of optima l ty regarding Quality of Service. The approach has been validated by a prototype and an evaluation in the case of an Health Information System. INTRODUCTION The number of available Web data sources and servic es has exploded during the last years. This enables users to access rich information in many do mains such as health, life sciences, law, geography, and many other domain of interest. Thank s to this wealth, users rely more on various digital tasks such as data retrieval from both publ ic and corporate data sources and data analysis with Web tools or services organized in complex wor kfl ws [Gao, 2005, Kinsi,2007]. However, human users have to spend uncountable hours to expl r and discover Web resources that meet their requirements. In addition, in many cases, use rs n ed to compose a specific set of Web resources in order to fulfill a complex question. T his situation is mainly due to the inability of present standards in capturing Web Service semantic s, i.e. the precise meaning of what a given Web Service exactly delivers regarding a specific u ser context. Meanwhile, Service-oriented computing (SoC) is emer ging as a new and promising computing paradigm that centers on the notion of service as t he fundamental element for accessing heterogeneous, rich and distributed resources in an teroperable way [Roman, 2005]. Web services are self-describing components that support a rapid and significant reuse of distributed applications. They are offered by service providers , which procure service implementation and maintenance, and supply service descriptions. Servi ce descriptions are used to advertise service capabilities, behavior, Quality of Service, etc. (U DDI, WSDL, OWL-S). Service descriptions are meant to be used by other applications (and possibl y other services), and not only by humans. WSDL and UDDI are the basic standards used for Web Service capabilities descriptions and advertising. However, they focus on the description of interfaces and syntactic considerations. So, at present, the development of powerful applica tions on the Web is still facing two major problems. The first one is related to the increasin g difficulties of identifying services that