Knowledge based Web Service Composition

Abstract

Automated Web service composition remains always one of the most challenging research area due to the continuously growing number of semantic Web services. Several formalization and composition approaches were proposed, but they are not mature and flexible. They assume that each sub-task is an atomic process, hence it cannot be decomposed based on user requirements. Moreover, these techniques and formalizations are not suitable for dynamic environments, does not consider incomplete and ambiguous query for which knowledge might be of great use. The aim of this work is to propose a novel horizontal and vertical composition of Web services based on Fuzzy Constraint Satisfaction Problem (FCSP) with knowledge integration. Knowledge, can be extracted from several sources, i.e. a datawarehouse of abstract workflows containing the best compositional practices in a given domain, ontologies, Web mining, etc. The proposed approach was implemented and applied to randomly generated examples using parameters reflecting the real world. Experimental results showed that the integration of knowledge with the CSP formalization gives remarkably better results in terms of the percentage of solved problems.