Research Fronts of Robust Social Recommendation

Abstract

Social recommendation plays an important role in solving the cold start problem in traditional recommender systems and improving the accuracy of recommendation results. However, it still faces serious challenges and problems due to bogus ratings or relationships injected by fake users. This practice seriously affects the authenticity of recommendation results as well as normal users' trustiness on recommender systems. Moreover, simply treating social relationships as a single type affects the recommendation accuracy. To solve these problems, in this paper, after summarizing the research of current social recommendation algorithms and detecting technologies of multiple relationships, we propose a framework for the research fronts of the robust social recommendation. This framework first models and extracts multidimensional relationships, then summarizes social recommendation shilling attack models based on the analysis of the relationships in social networks as well as the roles of relationships on recommendation. Finally, it forms robust social recommendation approaches that takes multidimensional relationships into consideration.