Dynamic Trustworthy Evaluation Approach For Web Services

In the past few years,the application of network technology has been becoming more and more mature.In this situation,Web services are developing rapidly because of open system and flexible structure.There're a large number of entities with similar functions and different attributes in the network environment.Networked software comprises a set of multi-source heterogeneous Web services widely distributed on the Internet.Under the influence of the dynamic and open network environment where the service is located,the credibility of the service is in constant change.Therefore,it's a highly urgent problem for the current industry to build a trustworthy evaluation method with dynamic adaptability for service selection and recommendation.Aiming at the problem that it's difficult to adapt evaluation to the real-time environment,we propose a dynamic credibility trustworthy evaluation method with Web services as the object in this thesis.Firstly,a dynamic weight self-tuning model for Web services credibility evaluation was described.In this way,the objective information in the fuzzy matrix is fully mined,and the subjective weights susceptible to human factors are moderately corrected by information entropy.A new relational matrix is formed by mining association rules between attributes.According to the different change types of the trustworthy attributes,the dynamic change process is analyzed deeply and quantified to form a dynamic adaptive weight reconstruction method.The method overcomes the inadequacy of the conventional subjective weight method which is inevitably affected by human factors.What's more,it ameliorates the problem of weak adaptability.To some extent,it provides a precondition for the dynamic adaptability of credibility evaluation method.In order to solve feedback evaluation problem of the Web services,a sort algorithm on the basis of lower bound of Bayes and updated evidence was proposed in this thesis.The Bayesian method is used to rank the favorable rating of the feedback from evaluation.Prior probability and return probability are introduced to retain the uncertainty.Prior belief is constantly updated through sampled evidence.In the next dynamic credibility trustworthy evaluation method,the feedback values are modified with a certain emphasis according to the time interval.At the end,a dynamic credibility trustworthy evaluation method was proposed in this thesis.Considering the influence of internal and external running environment on the current service credibility,we chose functionality,maintainability and safety as standard index,which is less associated with changing circumstances and can reflect credibility.Besides the evaluation of Web services by standard indexes,the modified indexes related to the management of quality control and external indexes are added.In order to improve the adaptability of the evaluation model,weights are calculated by the dynamic weight self-tuning model for Web services credibility evaluation.Dynamic indicators are composed of Bayesian inference feedback values by SALBEU algorithm.Finally,the analysis of experiment proves that this model is feasible in the credibility measurement.It could be able to adapt to the changing environment,which is capable of reconfiguring the service portfolio and overcoming the limitation in the previous credibility evaluation model.What's more,it's conducive to meeting the demands of users and evaluating the credibility of the whole system objectively.