Abstract:

The existing Point-of-Interest（POI） recommendation model based on user-based collaborative filtering considers two users have the same impact on each other. At the same time, only the user’s friend set is considered when calculating the similarity of social users, and the geographic information of user’s residence is not considered. In view of the above problems, this paper proposes a point of interest recommendation model which is Fuse User, Social and Geographical（FUSG）. Firstly, the asymmetric user impact and PageRank algorithm are integrated into the user-based collaborative filtering algorithm to mine the impact of user preferences on the POI recommendation system. Secondly, the similarity between users is calculated by combining the living distances and the common friends of social users, and the geographic features of user check-in are mined by using geographic information. Finally, the improved collaborative filtering algorithm, social information and geographic information are merged into FUSG model to recommend point-of-interest. The experimental results on real data sets show that the FUSG model not only alleviates the cold start problem, but also has higher recommendation results than other models.

Key words: points of interest, collaborative filtering, social information, geographic information, similarity

摘要：

目前基于用户的协同过滤兴趣点推荐模型认为两个用户之间对彼此的影响是相同的；同时，在计算社交用户相似度时仅仅考虑了用户的朋友集合，未考虑用户住所的地理信息。针对上述问题，提出了一种融合用户、社会和地理信息的兴趣点推荐（Fuse Users、Social and Geographic，FUSG）模型。将非对称用户影响和PageRank算法融入到基于用户的协同过滤算法中，挖掘用户偏好对兴趣点推荐系统的影响；结合社交用户之间的居住距离和用户的共同好友计算用户之间的相似度；利用地理信息挖掘用户签到的地理特征；将改进的协同过滤算法、社交信息与地理信息融合成FUSG模型，进行兴趣点推荐。在真实的数据集上的实验结果表明，FUSG模型不仅能够缓解冷启动问题，且与其他模型相比具有更高的推荐结果。

关键词: 兴趣点, 协同过滤, 社交信息, 地理信息, 相似度