Trust Evaluation in Unsupervised Network: A Fuzzy Logic Approach

Because of the possibility of anonymity and impersonation in social networks, trust plays an important role in these networks. In social networks, trust can have two aspects: trust of users to social network and trust of users to other users. Pear to pear networks, by eliminating the supervisor roles, besides its benefit in decreasing management costs, have problems in trust and security of users. In these networks, trust evaluation is only related to the trust of pear to other pear and because of the direct relation between pears; each user should know the trust level of other users. However, trust evaluation in pear-to-pear networks (as an unsupervised network), only can be done based on the past relation between pears or trust evaluation of other pears. This kind of trust evaluation cannot give a comprehensive view to pears. In other word, if any pear is not in the friend cycle of a user or friend cycle of user's friends, he will not be able to assign appropriate trust level to this pear. In this research, by using social networks as supervised networks, trust level of each user is evaluated, then by identifying these users in unsupervised networks, appropriate trust level is assigned to them.