Application of Services Relation Tracing to Automated Web Service Composition

Web service composition has received much interest due to its potential to achieve the visionary promise of Service-Oriented Computing paradigm in which services can be utilized as the basic constructs to create flexible dynamic business processes and agile applications that may span organizations and computing platforms with little effort even automatically. In this paper, we proposed a snowball method for automated Web service composition which is based on service relation tracing which starts off rather like a standard ego-centered network study. It discoveries one or several available Web services which can accept the input parameters in user requirement through sematic reasoning and determines whether its output parameters meet the user requirements. And then it fetches other candidate services following the Sequential-total and Sequential-partial relations, investigates their outputs, also asks further sequential relations, and so forth through a succession of "waves" of spreading. So the process is repeated and the composite service is constructed incrementally tracing sequential relations in Service Network just like a snowball. The experiment shows that the approach proposed in this paper can greatly improve the accuracy and efficiency of Web service composition.