Abstract: Web service composition has been increasingly challenging in recent years due to the escalating number of services and the diversity of task objectives. Despite many researches have already addressed the optimization of multiple Quality of Service(QoS) attributes, most of the currently available methods have to build a large web service dependency graph, which may incur excessive memory consumption and extreme inefficiency. To address these issues, we present a novel web service composition method by optimizing composition-segment candidates. Firstly, we formalize the web service composition problem as a Mixed-Integer Linear Programming(MILP) model and introduce some effective techniques for complex cases, and then a standard solver can be applied to this model. Afterwards, a candidate optimization method is proposed to solve the MILP model efficiently, which runs sharply fast without building a web service dependency graph. Experimental results on both Web Service Challenge 2009's datasets and substantial datasets randomly generated show that the proposed method outperforms the state-of-art while achieving a much ideal tradeoff among all the objectives with better performance.
Authors: Fang-Yuan Zuo and Yu-Bin Yang (Nanjing University, China)
Email: firstname.lastname@example.org, email@example.com