基于LBSN好友关系的个性化景点推荐方法

Personalized travel recommendation technology based on friendship of LBSN

基于位置的社交网络(LBSN)中照片带有丰富时间空间位置信息,为发掘用户偏好信息、进行景点推荐提供了条件。现有推荐方法存在推荐条件单一、难以准确估算用户偏好、推荐结果准确性不高的问题。改进传统协同过滤中相似用户计算和推荐方法,提出PTLR方法。通过用户景点照片矩阵计算用户偏好,结合好友亲密度信任关系计算相似邻居。利用多条件如兴趣偏好、景点时间适宜程度及候选周边关联景点产生推荐。实验结果表明PTLR能有效提高推荐准确性。

Photos in Location-Based Social Networks（LBSN）include rich time and space location information, which provides conditions for exploring user preference information and attraction recommendations. The existing recommenda-tion methods have many problems such as single recommended conditions, difficulty on estimation of user preference, low accuracy of recommendations. For those problems, PTLR method is proposed to improve the traditional method. Experimental results on LBSN real data show PTLR can effectively improve the recommendation accuracy.