Using multi-features to partition users for friends recommendation in location based social network

摘要: Abstract Friend recommendation is an important feature of social network applications to help people make new friends and expand their circles. However, the user-location user-user information in location based are both too sparse which contributes a big challenge for recommendation. In this paper, multi-feature SVM friend model (MF-SVM) proposed regarded as binary classification problem tackle challenge. We extract three features each user by methods respectively. The kernel density estimation entropy used smooth check-in data highlight activity level users spatial-temporal feature. Then extracted considering diversity common friends. After that topic improved LDA considers reviews corresponding service description textual Finally, these train whether have link can be predicted our model. experiments on real-world datasets demonstrate method paper outperforms state-of-art under different types evaluation metrics.