摘要

Open communities allow anonymous members to join and leave anytime, in which interactions commonly take place between parties who have insufficient information about each other. The openness brings risk to both the service provider and consumer. Trust is an important tool in human life, as it enables people to get along with strangers. It is accepted that trust management is a potential way to solve the above problem in computational systems. A trust management mechanism can assist members in the open community to evaluate each other and decide which and whether to interact. It provides an incentive to a good behavior and gives a punishment to a malicious behavior. The representation of trust in current computational trust and reputation models can be classified into four categories. We present a new evidential trust model based on the Dezert-Smarandache theory, which has a higher expressiveness than the trust models based on the Dempster-Shafer theory of evidence. We propose a smooth and effective approach of trust acquisition. The general rule of combination of the Dezert-Smarandache theory is used for trust aggregation. We consider that trust is transitive and present a rule of trust transitivity for our model. Lastly, we evaluate our model experimentally.