Abstract: By interpreting the common interest between the target user and his or her friends in social network into the latent factors based on the latent variable model, this paper proposes a Social Item Recommendation（SIR） model that encodes both the interest and relationship information such as friendship and friend-of-a-friend in social networks. It extends the SIR model to SIR+ by taking the social features into consideration, when making the inference of social item recommendation. The experimental results demonstrate that both SIR and SIR+ outperform the collaborative filtering methods, and the extended model SIR+ achieves a better performance than SIR model.

Key words: collaborative filtering, latent factor model, recommender system, social recommendation, social network

摘要： 将社交网络中目标用户和朋友之间相同兴趣产生的原因解释为潜在因子空间中的潜在因子，对社交网络中目标用户和朋友用户共同兴趣进行潜在因子分析，构建基于用户朋友关系的社交网络项目推荐模型，预测社交网络目标用户喜欢的项目。将基于社交网络项目推荐模型应用于实际应用场景中，研究表明与基于协同过滤技术的推荐方法相比较，该模型能够显著提高推荐质量，并具有良好的可扩展性。

关键词: 协同过滤, 潜在因子分析, 推荐系统, 社会推荐, 社交网络