Research On Automatic Web Service Composition Approach Based On Keyword Search

Service-based system(SBS)provides value-added composite services by composing existing loosely coupled Web services,which brings the large flexibility for the application development when the business changes frequently.Looking for the appropriate component services which are available for composition is one of the key steps in SBS project.The traditional Web service composition methods based on template customization require that system engineers have detailed field knowledge,which is not beneficial to achieve the agile SBS development;however,the dynamic Web service composition methods driven by the initial and target conditions cause that system engineers can not express their personalized functional requirements adequately.Under this background,this paper proposes the automatic Web service composition method based on the keyword search,the method allows system engineers search the service composition solution automatically by several keywords which can express the functional requirements under different constraint circumstances to achieve the quick construction of SBS.The main work and innovation points of this paper are indicated in the following aspects:(1)The semantic Web service model and the Web service composition problem integrating keyword are proposed.In particular,the Web service composition process is modeled as an AND/OR graph constructed by semantically matching input-output interfaces of Web services.On the basis of it,the initial service matching graph model and the service composition graph model are further proposed.(2)As for the general keyword search problems,the local keyword search algorithm based on Gaussian-weighted(LKS-GW for short)and the global keyword search algorithm based on number of services optimization(GKS-NSO for short)are proposed.By combining the above two algorithms,the service composition solution that satisfies the uniqueness of keywords with the minimized number of services can be extracted effectively in different scenes.(3)As for the keyword search problems with the Quality of Service(Qo S)constraints and the sequentiality constraints of keywords,the upper-bound pruning keyword search algorithm based on A* algorithm(KS-UPA* for short)is proposed.KS-UPA* adds Qo S decision-making in the search process,which can automatically filter the branch paths which don't satisfy the end-to-end Qo S constraints.At the same time,KS-UPA* combines the heuristic search of A* algorithm with the upper-bound global pruning strategy,which can quickly obtain the composition solution that satisfies the Qo S constraints with the minimized number of services.(4)Five Web service test sets with different scales are used to verify the effectiveness of the proposed algorithms.The evaluation results show that LKS-GW and GKS-NSO algorithms can generate a semantic input-output-based Web service composition efficiently that contains all the keywords queried by users while minimizing the number of services in the composition.In addition,KS-UPA* algorithm can generate a Web service composition quickly that satisfies the Qo S constraints,keyword uniqueness and sequentiality constraints.