Utilizing structural analysis of web of trust for evaluating the effect of deceptive data

Web of trust is the foundation of the reputation system, recommendation system and semantic web. Most of existing research on web of trust concentrates on aggregating the trust ratings on subjects and objects in the web of trust. The problem with this approach is that an adversary subject can accumulate reputation gradually and can be highly trusted by many other subjects. If this subject later deliberately releases a deceptive data, the effect caused by this deceptive data may be disastrous. Not only can this deceptive data greatly affect people who directly trust this individual, but also, its consequence may have an effect on many other subjects in the network. Our model illustrates how the structural analysis of web can help evaluate the deleterious result of the deceptive data.