A Enhanced Trust Model Based on Social Network and Online Behavior Analysis for Recommendation

The growth of social network applications makes it possible to incorporate relationship information to reason about trust value as many researchers have done with the trust-aware recommendation systems. But without explicit trust rating, inferring trust relationship among users become hard to accomplish. The contribution of this paper includes: 1) constructs user-oriented trust network based on small-worldness property; 2) enhances trust metric with considering social influence of the middle connectors and incorporates users potential relationship estimated from users' interactions frequency; 3) generate a compound weight to re-calculate similarity matrix combined with users social trust relationship.