Seminar: Preference-based Composition of Data-Providing Web Services

Graduate Program Seminar Announcement

Mehdi Razeghin
M.Sc. Candidate
Co-Supervisors: Dr. Adrian Fiech and Dr. Joerg Evermann

Preference-based Composition of Data-Providing Web Services

Department of Computer Science
Friday, December 14, 2012, 12:00 p.m., Room EN 2022




A Data-Providing (DP) Web service is defined as a software resource with the capability ofproducing particular outputs for certain given inputs. Recent years have witnessed a growing interestin consuming this class of Web services as a reliable method for exchanging data amongorganizations and individuals. In order to fulfill their requirements, these service consumers searchvia the Internet to find and invoke some available services. When no Web service is available tosatisfy their needs, a solution should be provided by combining several available Web services. Onthe other hand, as the number of available Web services increases, numerous composed services maybecome available. Hence, a ranking system is required to order the solutions according to users'preferences, i.e. inferior objectives that should be satisfied by a solution as much as possible. In thisthesis, DP services are modeled as non-recursive safe datalog rules, mediated ontologies areformalized by using full dependencies, and data integration techniques are applied to compose DPservices in the presence of these ontologies. We also proposed a novel algorithm by extending theMiniCon algorithm to compose Web services in the presence of full dependencies. Moreover, thecorrectness and the optimization of this algorithm are shown. Finally, we designed a formalframework to encode and utilize users’ preferences by which the best composed Web services canbe offered to them.