Trust Derivation and Recommendation Management in a Trust Model

Recommendation-based trust models have emerged as an important risk management mechanism in E-commerce and online environments. This paper presents a trust model with recommendations management based on subjective logic and Bayesian estimation. For dealing with recommendations, the theory of subjective logic is used to express trust properties as beliefs and a method is provided for mapping between similarity measures and trust. Using the derived trust in aggregation of recommendations, our proposal detects and protects against unfair recommendations. By using Bayesian estimation, our proposal integrates the subjective prior knowledge and the actual experience in to the estimation of trust. We present the test methods and their results to illustrate the accuracy and effectiveness of the approach.