Web service detection in service-oriented software development - a semantic syntactic approach

Service orientation has become a buzzword in research and practice during the last years. Enterprises consider investing in service-oriented architectures to increase the reuse of their existing software and to decrease development time and costs. On the one hand, standardization committees have advanced to define many layers of technical standards that are to become the technical backbone of a service-oriented infrastructure. On the other hand, the term service-oriented architecture is not precisely defined and a mature development methodology does not exist, as the numerous offers of software consulting services in this field show. This thesis makes a contribution to the ongoing discussion on the organization of service-oriented application development. First, the needs of a top-down development process for service-oriented architectures are examined by using a development scenario from financial industry. The ability to detect existing Web services during the design process is revealed to be the key to reusability. To support Web service detection, a relational Web service operation repository is introduced to store and search for existing Web services at the level of operations. The relational representation in the repository is then used to express the Web service detection problem as a relational schema matching problem on a syntactic level. The match between a Web service operation sought and an existing Web service operation is described with extended relational algebra expressions as a new transformation Web service. This representation is used to define a match hierarchy. The syntactic approach is extended to express semantics by adding semantic annotations to the relational Web service operation repository. The annotations are based on a taxonomy expressed in description logic. An algorithm is presented, that uses the taxonomy to rewrite SQL queries over the repository for operation retrieval. The algorithm supports the search for different match types. This approach combines Web service standards with semantic Web standards to provide structured algorithmic support for design time detection of Web service operations.