Friendship Link Recommendation Based on Content Structure Information

摘要

Intuitively, a friendship link between two users can be recommended based on the similarity of their generated text content or structure information. Although this problem has been extensively studied, the challenge of how to effectively incorporate the information from the social interaction and user generated content remains largely open. We propose a model (LRCS) to recommend user's potential friends by incorporating user's generated content and structure features. First, network users are clustered based on the similarity of user's interest and structural features. Users in the same cluster with the query user are considered as the candidate friends. Then, a weighted SimRank algorithm is proposed to recommend the most similar users as the friends. Experiments on two real-life datasets show the superiority of our approach.