用户关系和上下文感知的下一个兴趣点推荐

随着移动设备和社交软件的普遍应用，下一个兴趣点推荐（next POI recommendation）变成了基于位置的社交网络（LBSN）的一个非常重要的任务。现实生活中用户访问的下一个兴趣点通常受到用户签到序列信息、用户关系和该地点的上下文信息等诸多方面的影响。基于循环神经网络（RNN）的方法已经被广泛的应用到下一个兴趣点推荐中，但是这些基于RNN的方法缺乏对用户关系进行深入建模。为了解决上述问题，提出了一种整合用户关系和门控循环单元（GRU）进行下一个兴趣点推荐的模型（GRU-R），同时该模型能够考虑用户签到序列信息、用户关系、兴趣点的时空信息和类别信息等进行下一个兴趣点推荐。在两个真实公开的数据集上进行实验，结果表明提出的模型比现有主流的下一个兴趣点推荐算法具有更高的推荐准确性。

User Relationship and Context-Aware Next Point of Interest Recommendation

With the widespread application of mobile devices and social software, the next point of interest（POI） recommendation has become a very important task of location-based social network（LBSN）. In real life, the next POI is usually affected by the sequence information of user check-in, user relationship and the context information of the location. The methods based on recurrent neural network（RNN） have been widely used in the next POI recommendation, but these methods lack in-depth modeling of user relationship. To solve the above problems, this paper proposes a model（GRU-R） which integrates the user relationship and gated recurrent unit（GRU） to the next POI recommendation（GRU-R）. The proposed model can effectively recommend the next POI by considering user relationships, user history check-in sequence, spatiotemporal information and category information of visited POI. Experiments are conducted on two real public datasets. The experimental results show that the proposed model has higher recommendation accuracy than the existing mainstream next POI recommendation algorithms.