Research On Semantic Correlation Aware Composite Service Selection

With the widespread application and development of service computing technology, servicecorrelation is usually ignored in most current composite services selection methods. However, whensome services have semantic correlation, the actual running QoS performance of the compositeservices deviates from the declared QoS. Such that the final QoS values of composite services willdirectly affect user’s service selection result.Firstly, semantic correlation is employed to describe service correlation, and this paper analyzesand defines three semantic correlations which are control logic correlation(CLC), interfacedependent correlation(IDC) and business process correlation(BPC). Based on previous work, thispaper developes a series of correlation rules, and propose a multi-steps correlation mapping modelfor building related service resource pool.Secondly, this paper proposes a semantic correlation aware composite services selection method.A semantic-correlation aware QoS model for service selection is also presented taking semanticcorrelation between services as an important selection criterion. It searches and optimizes thecomposite services by iterative backtracking pruning function. Proposed services selection methodnot only effectively improves the QoS of composite service but also meets users’ requirments.Thirdly, this paper proposes an automatic composited service discovery framework which isused to support the discovery and reuse of composited service, it also proposes a temporal Petri-Netservice model(TPNSM) which is used to change the traditional business process model intotemporal Petri-Net service model. This paper also takes account of business process structure andinterface behavior’s impacts on the performance of composited service discovery. Proposedcomposited service discovery method not only can greatly improve recall rate and precision ofcomposited service discovery, but also can rapidly accelerate service selection process of compositeservices.Finally, based on the theory, the semantic correlation aware service selection simulation systemis designed and implemented. Simulation results demonstrate the feasibility and effectiveness of theproposed method.