Abstract:

The paper proposes an improved friend recommendation algorithm in social networks. It combines the relationship recommendation among friends and the similarity recommendation of tags. Firstly, through the friend relationship, the TopN common friends among the target users are recommended. Secondly, the tag similarity among the target users and the recommended TopN friends is used to recommend the friends with the highest similarity score, and they are given corresponding weights and scored.  Finally, the TopN uses with the highest scores are recommended. The results show that the proposed improved algorithm is efficient and its precision and recall rates are better than the common friend recommendation algorithms.

Key words: association rules;friendly recommendation;tag；similarity