Personalized friends recommendation is animportant wayto promote the service quality of social networks. In alarge-scale social network, accurately recommending friends with similar themes can make usersstickier, but the sparseness of the massive data makes most of the existing social networks cannot accurately make friend recommendations based on similarities in the interests of users. To address this issue, a personalized topic-oriented recommendation method based on user’s interest (ITOR) is proposed. Firstly, the topics of user’s interests are extracted by K-core analyzing method. Secondly,combining the users’ attributes information, the probability of being good friends can be computed. The satisfaction and accuracy of recommendation resultsare further enhanced with this probability. Finally, we verify the effectiveness of the proposed method by crawling SinaWeibo data released in September 2015.

 

personalized, friend recommendation, interest, topic similarity, prior probability