A Social Recommendation Algorithm with Trust and Distrust Considering Domain Relevance

Abstract

Although most of the existing social recommendation algorithms can alleviate data sparsity or cold-start problems, they only measure the influence of trust relationship on recommendation precision. To accurately measure the influence of the social relationship, we propose a social recommendation algorithm based on domain relevance and integrated into the trust and distrust relationship information. The algorithm is based on the Funk-SVD algorithm in which domain relevance of users is calculated by using cluster algorithm to find the groups, the trust and distrust relationship information of user are added, and the effect of user distrust on global influence is considered. Finally, Experiments based on the well-known Epinions show that the algorithm has obvious effects in improving the recommendation quality and alleviating the cold start problem.