测绘学报 ›› 2018, Vol. 47 ›› Issue (9): 1261-1269.doi: 10.11947/j.AGCS.2018.20170552

• 地图学与地理信息 • 上一篇    下一篇

关注点推荐算法的霍克斯过程法

张国明1,2, 王俊淑3,4, 江南3,4, 盛业华3,4   

  1. 1. 南京大学计算机科学与技术系, 江苏 南京 210023;
    2. 江苏省卫生统计信息中心, 江苏 南京 210008;
    3. 南京师范大学虚拟地理环境教育部重点实验室, 江苏 南京 210023;
    4. 江苏省地理信息资源开发与利用协同创新中心, 江苏 南京 210023
  • 收稿日期:2017-09-25 修回日期:2018-03-23 出版日期:2018-09-20 发布日期:2018-09-26
  • 通讯作者: 王俊淑 E-mail:jlsdwjs@126.com
  • 作者简介:张国明(1982-),男,博士生,工程师,研究方向为服务计算与大数据。E-mail:zgmming@qq.com
  • 基金资助:
    国家自然科学基金(41631175);江苏省自然科学基金(BK20171037);江苏省高校自然科学研究面上项目(17KJB420003)

A Point-of-interest Recommendation Method Based on Hawkes Process

ZHANG Guoming1,2, WANG Junshu3,4, JIANG Nan3,4, SHENG Yehua3,4   

  1. 1. Department of Computer Science and Technology, Nanjing University, Nanjing 210023, China;
    2. Health Statistics and Information Center of Jiangsu Province, Nanjing 210008, China;
    3. Key Laboratory for Virtual Geographic Environment, Ministry of Education, Nanjing Normal University, Nanjing 210023, China;
    4. Jiangsu Center for Collaborative Innovation in Geographical Information Resource Development and Application, Nanjing 210023, China
  • Received:2017-09-25 Revised:2018-03-23 Online:2018-09-20 Published:2018-09-26
  • Supported by:
    The National Natural Science Foundation of China (No. 41631175);The National Natural Science Foundation of Jiangsu Province (No. BK20171037);The Program of Natural Science Research of Jiangsu colleges and Universities (No. 17KJB170010)

摘要: 关注点(point-of-interest,POI)推荐是基于位置的社交网络(location-based social network,LBSN)中重要的个性化位置服务。针对LBSN中用户签到数据的复杂性和高度稀疏性问题,本文提出了一种基于霍克斯过程的上下文感知协同过滤关注点推荐算法(HWCF)。首先,根据用户签到关注点的地理空间聚集现象分析用户行为特征,筛选用户候选关注点;然后,利用霍克斯过程对候选关注点建模,通过融合空间距离信息、空间序列变换信息、时间信息、用户偏好、关注点流行度等多种上下文信息计算用户访问候选关注点的概率,对访问概率排序得到top-k推荐列表;最后,对算法参数的取值及调整过程进行讨论。试验结果表明,HWCF算法比其他的关注点推荐算法具有更好的推荐效果。

关键词: 关注点推荐, 基于位置的社交网络, 霍克斯过程

Abstract: Point-of-interest (POI) recommendation is a crucial personalized location service in LBSNs.To cope with the complexity and extreme sparsity of users check-in data,we proposed a context-aware collaborative filtering POI recommendation algorithm based on Hawkes process (HWCF).First,we analyzed users' behavior characteristics according to the geographic spatial clustering phenomenon of users' check-in POI,and filtered users' candidate POI.Then,we utilized Hawkes process to model candidate POI.Integrated different context information,such as spatial distance,spatial sequence transformation,temporal,users' preferences,POI popularity,etc.to compute the visiting probability of candidate POI for every user,and then obtained the top-k recommendation list by sorting the visiting probability.Finally,we discussed the range and adjustment of parameters in HWCF algorithm.Experimental results show that HWCF achieves better performance compared to other advanced POI recommendation algorithms.

Key words: point-of-interest recommendation, location based social network, Hawkes process

中图分类号: