Research And Application On Agent Based Web Service Composition

With the development of service-oriented computing theory and technology, the key technologies to the services---Web Services Composition (WSC) has become a heat point of the theory and application study. As its functions are limited, a single Web Service (WS) is unable to meet the actual demand, which means that it is necessary to compose multiple WSs to provide more powerful functions. After analysis of the existing methods of WSC, most of WSC approaches are centralized, which can not adapt to the needs of a distributed environment. Besides, WSC is mostly a static web service composition because of the passivity of the WS. Therefore, the dynamic and sovereign combination can not be achieved. These all bring challenges to the practical application of WSC.In order to reduce the pressure of WS being queried and improve its service utilization, this paper realized the conversion of WS from being called to searching the requirement by combining the Agent technology, and met the need of improving the autonomy and intelligence of it. The contents and innovations of the paper are highlighted as follows:(1) Web Service Compound Called Model (WSCCM)Compared with the traditional model of WSs, WSCCM has added one component, which is Demand Registry (DR). It is the basis of active service Agent searching for users'requirement. WSCCM not only provides the function of service requester querying WSs in the service registry, but also provides the function of active service Agent searching for users' requirement actively. Accordingly, it implements the two-way alternatives (the active and passive) between the requirements and WSs. And it meets the interactions and consultations between the service requester and service provider better.(2) Agent-based Modeling of Demand Registry (DR)The key elements in the DR are users' requirement and active service. We adopted Agent to model them, and discussed the key issues in the Agent-based DR model, especially, the needs decomposition of the need Agent and service capacity of active service Agent. The matching mechanism of user needs and service capacity is the key that active service Agent finds needs, which is the basis of completing service composition.(3) An Approach of active service Agent Finding Demand AgentBased on the Demand Agent and active service Agent model, according to the key issues in the process of active service Agent finding Demand Agent, the corresponding solution has been proposed. The first task is to solve the matching between the user needs and service capacity. On the basis of describing the matching mechanism of user needs and service capacity, the paper designs the matching strategy of user needs and service capacity.(4) (Particle Swarm Optimization) PSO-Based Dynamic Composition Service Selection AlgorithmBased on the active service Agent matching the user needs and the registration of service, in order to meet the non-functional needs of users, the PSO-Based dynamic composition service selection algorithm has been proposed. This paper not only describes the optimization process of dynamic services selection, but also verifies the validity of the algorithm by experiment.Based on the above contents, considering "the Bank Pledge Business of Steel Logistics" as the application background, this paper uses the above proposed method and technology to design and develop the prototype system in order to achieve the business functionality, and verifies the feasibility of the proposed model and effectiveness of the proposed algorithm.