Research On Dynamic Collaborative Combination Of Web Services

In the era of the rapid development of Internet technology and computer technology, both in academic and industry area, Web services and their composition technology have been developing rapidly, and widely acknowledged as an open service delivery. Web service has an excellent platform independence, interoperability and reusability, in order to meet users’needs, rapidly services recommended methods and services composition methods showed up. However, these services were only waiting for the process to play a role in the third-party call, but the increase in the number of services, makes a lot of long-term services are not called. Software Agent has the initiative, autonomy and evolutionary characteristics, it can obtain demand information of third parties, and send the party its own recommendation, Using the coordinate technology of Multi-Agent System which is consisting of independent Agent can achieve greater functionality. But it isn’t the same as platform-independent web services, do not have interoperability and flexibility. To make the service itself has intelligence, get smart web services and service composition algorithm can be coordinated and verify combination of efficiency and success rate of the service composition. To make the service itself has intelligence, get smart web services and service composition algorithm can be collaborated and verify combination of efficiency and success rate of the service composition. Composed the Agent and web services, service captured users’needs information in the network, service itself recommend their own to the requesters initiatively and in real-time. Abandon the platform Jack and Jadex which handling static requests, choose Jade platform to handle dynamic combination. A Service packaged by Agent uses preference based on off line partially observable Markov decision service composition algorithm (PBOL_POMDP) and heuristic online partially observable Markov decision service composition algorithm (HO_POMDP) complete the service composition process efficiently and independently, and the results are returned to service the caller. The results were solved with classical algorithms AEMS2and citations OPUMO1algorithm, and were verified through comparing the performance. Based on these, we did the following work:(1) Like web service extended by Agent, web service composition model in the Internet is extended to multi-Agent dynamic cooperative service composition system, summarize two functional structure based on dynamic coordinate of Multi-Agent Systems, given service composition framework that executed dynamic composition, compared between the traditional two-tier framework for dynamic multi-Agent framework and expanded two cooperative composition, composition of process is completely done by some of the services involved in the cooperative Agent, protecting efficiency and accuracy of service composition.(2) First, distinct basic services, structured and unstructured service. When using offline POMDP search algorithm composition strategy in global scope, the search is comprehensive but time-consuming. In this paper, a composition of process solved basic services to the structured service by PBOL_POMDP, the introduction of preference information in offline POMDP algorithm, both deals with the global problem of optimal planning, and increases the purpose of the search, which greatly reduces the time-consuming for searching, before the long term preferences and middle term preferences have not changed, the search results were saved as an important data.(3) Upon the traditional online POMDP algorithm, for solving composition of services with complex demand function, Solving the boundaries of the results of the off line algorithm as set of as heuristic information of an online algorithm, then further applied to the online algorithm. Using HO_POMDP services composition algorithms to solve the composition of all the services and infrastructure services generated between the front step.(4) Build a lab environment, the framework and the service composition process MADC_SYS simulated using JADE platform, packaged the simple service to deploy their own experiment, apply the extended offline and online POMDP algorithm to the dynamic process of cooperative composition of services, and effectively improve the success rate and time-consuming features of the service composition.