一种基于位置社交网络融合多种情景信息的兴趣点推荐模型

针对现有位置社交网络兴趣点推荐的研究工作主要集中在挖掘兴趣点的情景信息:时间信息、地理位置和评论信息，其中评论信息对用户偏好的影响尚未充分研究的情况。为此，提出一个统一兴趣点推荐模型，融合了用户偏好模型和上述3种情景信息，其中对用户偏好建模采用基于签到次数的度量方法，同时对评论信息采用基于潜在狄利克雷分配主题模型来挖掘用户偏好。实验结果表明，该模型在推荐准确率等多种评价指标上都取得了更好的结果。

UGTM: Exploiting various types of contextual information for Point-of-interest Recommendation on Location-Based Social Networks

Since the existing works of POI recommendation on location-based social networks (LBSNs) focus on mining context information of POI: the geographical information, comment information and the temporal information, which the comment information of user has not been systematically studied. This paper proposed a unified POI recommendation model, which fused user preference to a POI with temporal information, geographical influence and comment information of user. The framework studied the comment information of LBSNs by exploiting the Latent Dirichlet Allocation (LDA) model and modeled the user preference based on the number of user check-in behaviors. Finally, Experimental results in real world social network show that the proposed model outperforms state-of-the-art recommendation algorithms in terms of precision and rating error.