With the widely use of smart phones and social networks, service based on LBSN is becoming more and more important. Recommending a friend with similar interests for the registered users is one of the most important services of LBSN. Different from the traditional online social networks, LBSN integrates the user's online information and offline information, which greatly strengthens the interaction between the virtual world and the real world. In addition to the user profile
(such as age, gender, occupation, interests, etc.) and common friends, LBSN also records the history behavior of user check-in (such as check-in time and place), which suggests the user's real behavior patterns and consumption preferences. Research [1] shows that the physical location preference is positively correlated with the time and visit frequency, and research [2] suggests that more interaction in the real world can increase the probability of the formation of the friendship. Therefore, besides the similar user profiles and friendship in the virtual world, the offline behavior of users in the real world should also be considered to recommend friends for the user.