Abstract: Aiming at the safety problem of the web user information, proposed a subjective and objective method of user behavior trust evaluation model. Using the entropy-weight-method to calculate the objective weight and analytic-hierarchy-process to determine the subjective weight, then combining the previous two results to further calculate the integrated weights with optimal balance, also introduced long-term quantitative evaluation mechanism based on the Sliding Window. Finally, through the dimensionless treatment of various type evidences and get the final user-behavior trust evaluation value. The theoretical analysis and practical application show that the model can meet the subjective and objective relative balance optimization conditions, and can get more accurate user behavior trust evaluation value for each-time and long-time interaction. It lays the solid foundation of theory and practice for high reliable analysis of user behavior in key network application such as e-commerce and online bank.

Key words: trust evaluation of user behavior, integrated weight, sliding window