The Research And Application Of Network Trust Based On Fuzzy Support Vector Machine

The rapid popularization of the Internet brings convenience to our lives, but also brought some new problems. The Internet has the characteristics of anonymity and openness, that to say, anyone at any time and place can publish information or carry out a transaction through the Internet, without any identity information. This makes the lack of information for people on the internet to determine whether other people are malicious. Therefore, the study of trust system of the Internet has both theoretical value and practical significance.This dissertation conducts in-depth research on the trust and reputation models, and proposed a fuzzy support vector machine based trust model. Firstly the training data is extract from a social network through a series of algorithms, according to the characteristics of social networks. Then a multi-classification support vector machine VSM-FSVM is proposed, and it is used to learn the training samples. At last, we use the trained VSM-FSVM to calculate the trust of other samples. Simulation experiments show that the model is accurate, and have robustness in non-extreme conditions.Finally we developed a prototype system of Internet Trusted Service Support System. This system collects data from Internet entities, and calculates static and dynamic credit for Internet entities, so it can provides credit protection for Internet users.