Quantitative Analysis of Trust Evidence in Internet

With the development of the computer network, Internet environment is increasingly complicated and unpredictable, which makes more applications need to establish trust between two entities, especially among strange entities. The trust includes direct trust, indirect trust and reputation trust. Indirect trust and reputation trust are both based on direct trust. Meanwhile, the direct trust is based on trust evidences obtained from past actions during interaction with other entity. The paper puts forward a method of how to update trust evidences dynamically and how to use updated trust evidence to compute trust and trust attribute of evaluated entity. We have proved that the method have good trust characteristics, such as "rise-slowly, decline-quickly", which can prevent from attack of malice entity; the subjective trust factor beta embodies subjective characteristic of updating trust evidences. With the increase of the number of interaction, the method can automatically improve trustworthiness of the evaluated entity, so the algorithm is self-adaptive. Meanwhile, when the entities do not interact with each other for a long time, the evidence value would decrease gradually and automatically; the method can prolong updating time automatically by computing dynamic standard variance of trust evidence, which can improve performance.