Traditional social recommendation systems have a major weakness: these recommendation systems can be easily affected or cheated from malicious users who create ad hoc user profiles and provide fake ratings and reviews. In such a situation, the recommendation systems cannot be reliable. For example, astroturfers are hired by some movie companies to distort certain movie’s ratings on IMDB. This project aims at exploring novel ways on how to prevent this situation and provide more trustworthy recommendations by systematically exploiting the multi-layer trust relationships in social media.