A method for evaluating initial trust value of direct trust and recommender trust

This paper aims to get initial trust value of direct trust and recommender trust. Firstly, we gives domain representations for trust value and trust information structure; then defines trust purpose as class with properties, computes subordinate degree of each property to each trust rank based on fuzzy sets. For evaluating initial trust value of direct trust, we calculate correlation coefficient; establish a reasonable basic probability assignment for each property; merge basic probability assignment functions of all properties using evidence combination rule to get basic probability assignment function of trust purpose as the initial value of trust value. The calculation of initial trust value of recommender trust is similar to the calculation of direct trust value; the difference is that with the similarity of properties instead of the correlation coefficient of properties to construct basic probability assignment function. After getting trust value vector of direct trust and recommender trust, we adopt the decision-making based on the basic probability assignment to obtain the final result. The method that integrated processing of all properties using evidence theory gets trust value of trust purpose can eliminate the redundancy and conflicts that may exist between properties, can reduce uncertainty, obtains more accurate and reliable conclusions, produces meaningful information of trust purpose. The method that obtaining the final result with a certain decision-making selection rules after achieving trust vector on all trust sets provides a more rational approach for the study of trust level evaluation problem.