融合多种影响因子的兴趣点推荐算法

为了解决兴趣点推荐任务中的数据稀疏性问题和充分利用位置社交网络中的多样信息提高个性化推荐质量，提出了一种融合多种影响因子的兴趣点推荐算法。分别对地理信息和社会信息进行地理影响力建模和社会影响力建模，并联合时间信息和地理信息进行时空影响力建模，然后以加权求和的方式整合3种影响力评分得到用户偏好分数，根据用户偏好分数为每个用户提供1个包含Top-NWT]个兴趣点的推荐列表。实验结果显示，在2个公开数据集上，融合多种影响因子的兴趣点推荐模型的性能优于对比模型。地理-社会-时空影响是兴趣点推荐任务中的关键，对这3种影响建模可为融合关键信息的兴趣点推荐研究提供参考。

Point-of-interest recommendation algorithm integrating multiple impact factors

In order to solve the problem of data sparseness in the task of point-of-interest recommendation and make full use of the diverse information in the location-based social network to further improve the quality of personalized recommendation, a point-of-interest recommendation algorithm integrating multiple impact factors was proposed. Geographic influence modeling and social influence modeling were performed on geographic information and social information, and temporal information and geographic information were combined to model temporal and spatial influence, and the three influence scores were integrated in a weighted summation manner to obtain user preference score. According to the user preference score, each user was provided with a recommendation list containing Top-N points of interest. The experimental results show that on the two public datasets, the point-of-interest recommendation model that integrates multiple impact factors performs better than the baselines. In addition to the user check-in frequency, the geographic-social-spatial-temporal influence is also a key part of the point-of-interest recommendation task, and the modeling of these three influences is of great significance, which provides certain reference value for the research of point-of-interest recommendation integrating key information.