Research On Reputation-based Web Service Selection

With the increase of web services in the number, It is inevitable that several candidate services having the same function coexist in the service discovery process. How to select the most trustful service that satisfies the users'expectations has become a key issue, which is technological problem that this thesis emphatically researches on.Traditional service discovery methods are generally executed according to service functions, service selection inevitable has the problem of blindness and randomness. At present, many service discovery methods are non-functional description based on quality of service. However, the quality of service property information often comes from service providers themselves, which does not take into account the dynamic changes of the quality of service completely, but also easily produce some malicious behavior such as chief. So, a reliable, trustworthy service selection methods is expected.In this thesis, the starting point is web service discovery and the semantic web services technology, the main study is how to select the trustworthy service in the distributed environment. The main research work include the followings:(1) Learn from the laws of social behavior of human, we discuss the web services based on trust and reputation. On the basis of existing trust model, we present a trusted web services model based onβprobability density function, and introduced the specific implementation method of the model. The model makes use of Bayesian statistical analysis functions which can predict the service's future behavior based on it's past behavior. The trust evaluation system that we has constrcted can improve the accuracy of services'selection.(2) The process of web services selection is divided into two stages, the match based on the semantic similarity ensures a high recall ratio, and the exact match based on the reputation can locate to the best services that the user expectations, the precision rate can be improved also. Two stages of implementation of the specific process are described and the service selection algorithm are designed.(3) The example in this thesis illustrates the selection process of web services based on reputation, a phased approach of service discovery ways make services search performance improved. Simulation tests prove that the method proposed in this thesis has a very good inhibition on the malicious behavior existing on the network, thus it ensures the selection of services with high credibility.