Research On Qos-oriented Trustworthiness Evaluation Approach For WEB Service

With the gradual increase of the web services with the same or similar function on the Internet, users'demands for quality of service are increasingly heightened. In practical use, the uncertainty of many factors such as service providers, service users, quality of service and other factors, can cause a variety of trustworthiness problems in the distributed, loosely coupled, flexible and dynamic web services environment. This is not conducive to searching and choosing the service which has high quality of service for users, so that it is difficult to ensure its own healthy operation and well development.By analyzing the characteristics of QoS and the impact of context factors on the results of web service QoS evaluation, this paper proposes the idea that, based on the difference of subjective and objective elements, the QoS attributes are classified, and through the in-depth research in the different types of QoS, proposes a context-aware oriented web service QoS prediction model, which is easy to extend.At the same time, for the issues in the trustworthiness evaluation approach for Web Services, this paper has an in-depth study. Firstly, for the issue that context factors directly affect the evaluation results of QoS, this paper presents the ontext-aware oriented weight learning algorithm to determine the weights of context factors associated with QoS attributes, in this way, context is successfully integrated into the QoS prediction; Secondly, taking into account the issue that different types of QoS have different subjective and objective elements, proposes a similarity calculation method based on the QoS classification, on this basis, proposes a context-aware oriented web service QoS prediction algorithm; Finally, for the characteristics of the malicious data, focus on fundamentally enhancing the resistance capability of evaluation algorithm, by quantifying the deviation between the malicious users'feedback value and the overall feedback value, a resistance capability enhanced trustworthiness evaluation algorithm for web services is proposed.To verify the validity of these findings, this paper completed comparative experiments between their method proposed in this paper and some existing methods programming with java. Through a large number of experiments under different conditions of users'number, some problems in this method are summarized, which provides direct and valuable experience for further study.