A Trust-Based Clustering Approach for Identifying Grey Sheep Users

摘要

In the context of recommender systems, users may have particular tastes and very unusual preferences comparing to the others. These users are called Grey Sheep Users. It is difficult to find similar users and relevant recommendations for such kind of users. In this work, we deal with trust values in learning users' behaviours and relations between each other. A trust-based clustering approach is proposed for identifying Grey Sheep Users. The obtained cluster is then exploited to make recommendations for target unusual users.