基于位置社交网络的兴趣点推荐研究  

Survey of research on point-of-interest recommendation methods based on location-based social networks

作　　者：李征[1,2,3] 黄雪原 袁科[1] Li Zheng;Huang Xueyuan;Yuan Ke(College of Computer&Information Engineering,Henan University,Kaifeng Henan 475004,China;Henan Key Laboratory of Big Data Analysis&Processing,Henan University,Kaifeng Henan 475004,China;Henan Spatial Information Processing Engineering Laboratory,Henan University,Kaifeng Henan 475004,China)

机构地区：[1]河南大学计算机与信息工程学院,河南开封475004 [2]河南大学河南省大数据分析与处理重点实验室,河南开封475004 [3]河南大学河南省空间信息处理工程实验室,河南开封475004

出　　处：《计算机应用研究》2022年第11期3211-3219,共9页Application Research of Computers

基　　金：国家自然科学基金资助项目(61402150,61806074);河南省科技攻关计划资助项目(182102410063);河南省高等学校重点科研项目计划(23A520016)。

摘　　要：兴趣点推荐是基于位置社交网络中的研究热点之一。首先对从Web of Science收集的兴趣点推荐研究文献进行了分析;然后分析了影响兴趣点推荐的多种因素,并在分析传统兴趣点推荐方法基础上重点从用户历史签到信息建模和用户社交生成信息提取两个方面对基于深度学习的兴趣点推荐方法进行了分析;最后,对未来可能提高兴趣点推荐效果的研究方向进行了展望。Point of interest recommendation is one of the research hotspots in location-based social networks.Firstly,this paper analyzed the references about point-of-interest recommendation research from Web of science,then analyzed the factors that influenced point-of-interest recommendation.Furthermore,based on the analysis of traditional methods of point-of-interest recommendation,this paper mainly analyzed the methods of point-of-interest recommendation based on deep learning from two aspects:modeling of user historical check-in information and extraction of user social generated information.Finally,it looked forward to future research directions that might improve the performance of point-of-interest recommendation.

关 键 词：兴趣点推荐 数据稀疏 访问序列模式 注意力机制 图嵌入 

分 类 号：TP311[自动化与计算机技术—计算机软件与理论]