Transaction similarity-based contextual trust evaluation in e-commerce and e-service environments

The trust of sellers and transactions is a very important issue in e-commerce and e-service environments. At some e-commerce websites (such as eBay¹), the trust management mechanism can compute a trust value of a seller, which is based on the ratings of past transactions given by buyers. This trust value, however, is static and can only reflect the general or global trust status of a seller, and it is not directly bound to a new transaction. As a result, a buyer may be easily cheated by a malicious seller in a new transaction with the notorious value imbalance problem [5], i.e., the malicious seller can build up a good reputation by selling cheap products/services and then start to cheat buyers by selling expensive products/services. Instead of providing such a static trust value, in order to provide more objective trust result for a new potential transaction, a trust evaluation mechanism should based on the ratings of past transactions, the nature of both past transactions and the new transaction. In this paper, we propose a new contextual trust evaluation method. Our method compares the transaction context similarity between the new transaction and past transactions, from which the trust value of the new transaction can be determined. Our method can identify and prevent potentially malicious transactions with the value imbalance problem.