The Research Of Web Service Composition Techniques Base On Multi-Agent

With the rapid development of computer network, the number of web services increases drastically. In an indivisuation oriented world, it is very important to satisfy the proprietary requirements of specific customer and enhance the usage rating of web service.Web Service Composition (WSC) is focused on Sementic Web, which is a relative professional tool and hard for users to master. For web services under different semantic environment, it is hard for them to communicate, and efficient mapping technology is still lack for communication of heterogenous data. For WSC, the ubiquity and adaptability is of high importance.Agent belongs to category of reusable software. Agent utilizes the fundamental services provided by system software, connecting different parts of complex webservice or different applications, can achieve the sharing of resource of different web services. In a word, calculation, storage and information resource can be shared among web services. Using Agents in WSC, Intelligence of Agent can be exploited and efficent management of web service can be accomplished. It is still in a exploring stage for application of Artificial Intelligence Algorithm in WSC. Particale Swarm Optimazation (PSO) has advantages of fewness of units, computaion simplicity, good robustness and etc, which is excel at optimization for versatile multi-dimentions consecutive space. PSO is fast in convergence but lacks the ability of mutation and prones to radiation.In this paper, the background of Web Service Composition techniques, definition, infrastructure and characteristics of Agent are discussed firstly, and in addition, existed technologies are analyzed and summarized. On the basis of analyzing deficiencies of current Agents infrastructure, a multi-Agents framework based on software bus and layered management model is proposed.On the basis of analysis and comparision of current WSC algorithms, WSC is treated as a MOP (Multi-Optimization Problem), and a pertinent Hybrid Partical Swarm Optimation and Genetic Algorithm (HPSOGA) are proposed. GA is used for global searching and PSO is used for enhancing local searching. The description, convergence demonstration is proposed. TSP is used for algorithm performance comparision.At last, HPSOGA is implemented in an e-commerce simulation system, namingly VMART. The experiment shows the excellent performance of the algorithm, and the usability of the algorithm is demostrated in elementary stage.