PSO-inspired BIRCH and Improved Bipartite Graph for Automatic Web Service Composition

Web services are published by the service providers through the internet as independent software components by fulfilling the requirements of the customer requests. As the customer requests are not always satisfied by a single service, the methods of services composition were evolved in which a chain of services are composed together. However the longer search time for the discovery of composed services is a challenging problem causing the customer to wait long time especially while using larger web service set. In this paper, an efficient automatic web service composition approach is proposed to improve the performance by introducing domainontology-based PSO-inspired Balanced Iterative Reducing and Clustering using Hierarchies (BIRCH) clustering and Improved Bipartite graph in the pre-processing and service discovery & ranking phases respectively. Domain-ontologybased PSO-inspired BIRCH incrementally and dynamically clusters the web services in hierarchical manner with optimizing the clustering results to reduce the processing time. The approach improves clustering quality by additional scans and thus creating sophisticated environment for service matchmaking. In discovery phase, the Improved Bipartite graph produces maximal bicliques by pruning the non-maximal bicliques in the search space without restricting the inputs and reduced computations for elimination. Thus the best service clusters is discovered and ranked which ensures the selection of best composition plan in the planning, verification & execution phase. Experimental results performed in terms of accuracy, runtime, recall and precision, shows that the proposed framework improves the efficiency of web service composition.