一种融合用户空间行为特征的兴趣点推荐算法  

A POI Recommendation Algorithm Integrating User's Space Behavior Characteristics

作　　者：李华孝杨 徐青[1] 王卓苧 朱新铭 黄文君 LI-HUA Xiaoyang;XU Qing;WANG Zhuoning;ZHU Xinming;HUANG Wenjun(Information Engineering University,Zhengzhou 450001,China;61206 Troops,Beijing 100043,China)

机构地区：[1]信息工程大学,河南郑州450001 [2]61206部队,北京100043

出　　处：《测绘科学技术学报》2024年第6期654-660,共7页Journal of Geomatics Science and Technology

基　　金：国家自然科学基金项目(42101454)。

摘　　要：在综合考虑用户签到数据特点和密度聚类算法优势的基础上,提出一种融合用户空间行为特征的兴趣点推荐算法。首先,通过用户活动能力分析,剔除签到噪声点,构建用户签到倾向—签到点间距关系模型;其次,针对不同用户采用KANN-DBSCAN聚类算法分析用户活动区域,捕获用户空间分布特征并融合用户空间行为特征对Top-N兴趣点进行推荐;最后,采用Gowalla数据集对本文算法与其他5种算法进行实验对比,并通过准确率和召回率两个评价指标验证本文算法的有效性。实验结果表明,该算法有效提高了兴趣点推荐的质量。In this paper,a POI recommendation algorithm is proposed in which user's spatial behavior characteris⁃tics are integrated based on a comprehensive consideration of the characteristics of user's check⁃in data and density clustering algorithms.Firstly,through the analysis of user's activity ability,the noise points of sign⁃in data are e⁃liminated,and the relationship model between user's sign⁃in tendency and sign⁃in point spacing is built.Secondly,the KANN⁃DBSCAN clustering algorithm is used to analyze the user's activity area for different users,and to cap⁃ture the user's space distribution characteristics.The user's space behavioral characteristics are integrated to rec⁃ommend Top⁃N POI.Finally,Gowalla data set is used to compare the proposed algorithm with the other five algo⁃rithms,and the effectiveness of the algorithm in this paper is verified by two evaluation indicators of the accuracy and recall.Experimental results show that the proposed algorithm effectively improves the quality of POI recommen⁃dation.

关 键 词：兴趣点推荐 空间行为特征 聚类 活动能力 活动区域 

分 类 号：P208[天文地球—地图制图学与地理信息工程] TP311[天文地球—测绘科学与技术]