Abstract: There are diffrent qualities of Web services in Web environment. However, how to choose reliable Web service that satisfies users' requirements has become a serious problem. We proposed a trust mechanism based on requirement contexts and recommendation dishonesty. Filter unvaluable historical data by calculating requirement contexts similarity and find the appropriate recommenders by fuzzy clustering of recommenders' different preferences. Then we proposed a approach based on recommendation dishonesty and honesty to distinguish recommenders' capacity to weaken malicious recommenders' influence. At last, the results of experiment demonstrate the feasibility and validity of the approach.

Key words: Webservice quality, Requirement contexts, Fuzzy cluster, Recommendation dishonesty