Can reputation migrate? On the propagation of reputation in multi-context communities

摘要

As e-communities grow in both quality and quantity, their online users require more appropriate tools to suite their needs in such environments. Many such tools are not explicitly needed in real-world communities where humans directly interact with each other. Trust making and reputation ascription are among the most important examples of such tools. Humans often build trust relationships through interaction or recommendation, and are therefore able to ascribe relevant reputation to those they interact with. However, in online communities the process of trust making and reputation ascription is more complicated. In this paper, we address a special case of the trust making process where community users need to create bonds with those they have not encountered before. This is a common situation in websites such as amazon.com, ebay.com, epionions.com and many others. The model we propose is able to estimate the possible reputation of a given identity in a any new context by observing his/her behavior in other communities. Our proposed model employs Dempster–Shafer based valuation networks to develop a global reputation structure and performs a belief propagation technique to infer contextual reputation values. The preliminary evaluation of the proposed model on a dataset collected from epinions.com shows promising results.