Research Of Web Service Composition Based On Integratedly Consideration Of Functional And Nonfunctional Attributes

With the rapid development of e-commerce and Web services gradually increasing, Web services have been widely recognized as an effective mechanism for data and information integration on the web. Nowadays, how to quickly and accurately gain and combine massive Web service into a new and complex service in order to meet the need of users’functional and nonfunctional attributes has been a hot topic on Web Service Composition. At present, the research on Web Service Composition still exist some defects such as imbalanced load, inefficiency and improper composition methods and so on. The major tasks done in this thesis for the above problems are:Firstly, constructing the Web service composition framework based on feedback mechanism. A new Web service composition framework based on feedback mechanism has been built by adding a feedback model on the basis of the existing Web service composition framework. The feedback model can not only timely feed back the status of service composition in order to resolve the problems of load balancing and changes of QoS values, but also can feed back the evaluation of the service requester, which can provide reference to other requesters’access.Secondly, establishing a candidate set of service generating framework with functional attributes. When considering the match of the functional attributes, namely in the generation process of selecting the candidate set of services, the index mechanism which is based on users’comments and the click rate is added to the framework, the updating mechanism of the index is designed, the evaluation function of the index is constructed and a candidate set of services generating framework with function of index mechanism is proposed in order to improve the efficiency of the functional match.Thirdly, proposing Web service for global optimization algorithm with nonfunctional attributes. When considering selection of non-functional attributes, which is also called generating composition set of service of service selection, in order to quickly and accurately select Web service to meet users’ need of nonfunctional attributes from massive Web service of the same or similar functions, a selection algorithm of Web service based on the improvement of the ant colony algorithm is proposed. It introduces chaotic model when ant colony initializing, and adds chaotic disturbance when updating the pheromone. On the one hand, by using the pheromone positive feedback of the ant colony algorithm, it helps strengthen the chaos algorithm in the search area, increase the number of times of searching, and reduce the blindness of the chaos algorithm when it’s used in searching; On the other hand, by using the initialization of Chaos colony algorithm, it can also improve the diversity of individuals and avoid premature.Last, designing and implementing the prototype of Web service composition system. On the basis of the analysis of existing Web service discovery framework, the prototype system is designed and implemented. It concludes three parts including service querying, discovery and combination, which verifies effectiveness and feasibility of method of candidate set of services generating framework with functional attributes and Web service of global optimization algorithm with nonfunctional attributes.