A recommendation-based reputation evaluating model using virtual verifying and time decaying for P2P networks

As a recommender in the reputation systems can submit inaccurate recommendations for a variety of reasons, the trustworthiness degree of recommending (TDR) must be critically assessed. In this article, two mechanisms, i.e. the virtual verifying and time decaying, are introduced to help calculating TDR. The former is aimed to solve the accuracy and efficiency problem caused by insufficient evidences of recommending when calculating. The latter focuses on detecting the oscillation of recommender's behaviors, and time-decaying-rate factor and time-decaying-frequency factor are involved to make the recent evidences be more important to the calculating. Experimental results show that our model has better performance and is robust even with large amount of malicious peers.