Research On The Method And Technique Of Web Services Composition Based On QoS Constraints

In order to improve the interoperability of various applications from different platforms and operating systems, service-oriented computing model springs up during the past decade. As the core technique of SOC, web service technology has been studied widely. In the research area of web service technology, web service composition is one the most important research topic and has drawn the attention of many researchers. Web services composition is technology that reuses and aggregates existing web services to reduce software development costs. With the number of web services increasing, QoS-aware web services composition becomes a hot research topic. Its objective is to find a solution for web services composition problem, and the solution maximizes an application-specific utility function under QoS constraints.In view of that, the focus of the paper is the research onmethods and techniques of web services composition based on QoS constraints. A local optimization method satisfying global constraints and a decision support technique for uncertain QoS constraints (PQC) are proposed in the paper.The local optimization method satisfying global constraints utilizes tasks’weights to overcome the defect of the traditional ordinary local optimization methods which could not satisfy global QoS constraints. The method utilizes two hierarchical weights, i.e., the weights of tasks and the weights of task’s criteria, to reflect customers’ preferences, so that it improves the satisfaction of customers. The process of the method is composed of three steps:task weight calculation, global constraint decomposition, local optimized selection.The decision support technique for uncertain QoS constraints (PQC) predicts client’s uncertain QoS constraints based on collaborative filtering technique. PQC improves the friendliness of the web services composite system through helping clients to make a decision for uncertain QoS constraints. PQC tool is developed based on PQC technique. The technique consists of four steps. Firstly, QoS data set is built up. Secondly, all of QoS data are scaled. Thirdly, the dimensionless values of uncertain QoS constraints are predicted based on Slope One method. Then, the calculated dimensionless QoS constraints are reverted to dimensional values.Finally, it uses instances to verify the correctness, reasonableness and operability of the method and technique above. It uses an online shopping for T-shirt scenario to validate the local optimization method clearly. And it uses another online shopping scenario for flowers scenario to demonstrate PQC technique distinctly.