【摘要】 随着基于地理位置的社交网络的兴起,兴趣点(POI)推荐引起了人们的许多关注。POI推荐向用户推荐他们可能感兴趣但没有访问过的地方,从而解决用户"下一步去哪"的问题。本文提出新的用户相似性度量、全局影响力以及热门POI的概念。综合考虑了多个影响因素之间的关系,以地理分层结构的矩阵分解模型(HGMF)为基础,提出新的POI推荐算法HGS-MF。在Yelp和Gowalla社交网络数据集上对HGS-MF进行了评估。实验结果表明,HGS-MF方法的实验表现均优于传统的POI推荐算法。更多还原

【Abstract】 With the rise of location-based social networks,Points of Interest(POI) recommendations have attracted a lot of attention.The POI recommends places which users may be interested in but have not visited,thus addressing the user’s issue about Where to Go Next.This paper proposes new user similarity measures,global influences,and the concept of popular POIs.Based on the relationship between multiple influencing factors,a new POI recommendation algorithm HGSMF is proposed based on the Hierarchical Geographical Matrix Factorization model(HGMF).HGS-MF is evaluated on the social network datasets of Yelp and Gowalla.The experimental results show that the performance of the HGS-MF method is superior to that of the traditional POI recommendation algorithm.更多还原