Personalized Composition Of Web Services Based On Customer Satisfaction

In recent years, E-commerce has evolved from localization to globalization, from B2C (Business-to-Customer) pattern to B2B (Business-to-Business) pattern, and from the centralized model to the distributed model. As a fresh pattern of Web applications, Web service is a new distributed computing paradigm and an effective mechanism for data integration and information integration. Since a single and simple Web service cannot meet the needs of practical applications, we need to combine multiple Web services to accomplish a complex task. Service composition should take into account a customer's requirement and interest, and then combine those services to satisfy the customer's integrated service requirement.Personalized services are those which supply different service strategies, contents and models according to a customer's characteristic. To establish a personalized service combination means that one has to combine the existing services and make them work corporately to meet the requirements of users. However, most current personalized Web services combinations only take technical realization into account but failed to consider customers'degree of satisfaction from the perspective of business. Therefore, they cannot completely meet the personalized requirements of customers.To solve the problems mentioned above, the thesis carries out an in-depth research on Web services combination system from the following perspectives:1. Considering the problem related to customers in the current personalized services combination systems, we have added the important factors of service customer satisfaction and applied it to the combination of personalized services.2. By considering customers'degree of satisfaction as an attribution of services combination case, one can find the alternative combination solutions based on Case-based Reasoning (CBR) technology. In the case-retrieval process, a domain ontology-based similarity algorithm is designed which brings up an attribution conceptual semantics based similarity algorithm.3. Considering customer satisfaction, we establish a weight vector of attributes preferred solution model and find the optimal solution from the candidate solutions.4. An intelligent, reactive, and proactive multi-agent technology is applied in the Web service composition system to provide powerful technical support for information and resource discovery and combination. A multi-agent combination system-based overall architecture model—CMSCS (CBR-based Multi-Agent Web Services Composition System) and its function, structure and workflow are brought up. CMSCS is composed of five kinds of independent agents which independently and corporately to complete the combination task, getting rid of the limitation of the traditional combination system.5. On a basis of the above technology and algorithm, this thesis preliminarily establishes a web service combination system for educational resources system (SCSRS), and designs corresponding experiment to test the system, the results of which show the effectiveness and practicality of the algorithm.In summary, this thesis integrates CBR method and Multi-agent technology to the application of personalized Web service combination system on the basis of customer satisfaction, and makes a useful attempt in the Multi-agent collaborative technologies.