摘要
当前兴趣点推荐大多利用兴趣点的位置信息和用户的社交关系提升推荐质量,忽略了兴趣点评论信息的重要性;此外,推荐的兴趣点之间通常比较相似,不具有代表性和差异性.针对上述问题,提出了一种新的兴趣点相关度评估模型,称为地理-社会-评论关系模型,并给出了一种新的评论文本相似度度量方法.根据兴趣点间的地理-社会-评论关系相关度,提出了基于谱聚类的兴趣点聚类方法和基于概率密度估计的兴趣点典型化选取方法,以便从每个聚类中选取一个具有代表性的兴趣点.对于选取的典型化兴趣点,提出了利用概率因子模型拟合用户访问兴趣点次数矩阵的方法对推荐结果进行个性化排序.实验结果表明,本文提出的相关度评估模型对兴趣点的相关度评估更合理,推荐结果在多样性和准确率方面都取得了更好的效果.
The existing point-of-interest( POI) recommendation methods mainly use the geographic information of the POI and the social relationship of the users to improve the recommendation quality while the influence of the POI’s comments are neglected. Furthermore,the result POIs in recommendation list are usually similar to each other which makes them losing diversity. To deal with the problem mentioned above,this paper builds a new POI relationship measuring model,which is called Geo-Social-Comment relationship model. For understanding the comments,a new method to measure the similarities of comment texts is also proposed. According to the correlations between POIs,a spectral clustering-based POI clustering method is proposed and then a probability density-based typicality selection method is presented in order to pick the representative POI from each cluster. For the selected typicality POIs,a personalized ranking method,which leverages the probabilistic factor model fitting the user accessing matrix,is presented. Experimental results demonstrate that the proposed POI relationship measuring model is more reasonable to evaluate the relevancy of POI,and the recommended results have achieved better efficiency in diversity and accuracy.
作者
孟祥福
毛月
张霄雁
赵路路
赵泽祺
MENG Xiang-fu;MAO Yue;ZHANG Xiao-yan;ZHAO Lu-lu;ZHAO Ze-qi(School of Electronic and Information Engineering,Liaoning Technical University,Huludao 125105,China)
出处
《小型微型计算机系统》
CSCD
北大核心
2019年第11期2431-2438,共8页
Journal of Chinese Computer Systems
基金
国家自然科学基金项目(61772249)资助
辽宁省自然科学基金项目(20170540418)资助
辽宁省教育厅项目(LJYL018)资助．
关键词
兴趣点推荐
地理-社会-评论关系模型
谱聚类
典型化选取
POI recommendation
geo-social-comment relationship model
spectral clustering
typically select
作者简介
孟祥福,男,1981年生,博士,教授,博士生导师,CCF会员,研究方向为空间数据管理、推荐系统、Web数据库查询.E-mail:maoy27@126.com;毛月,女,1994年生,硕士研究生,CCF会员,研究方向为短文本分析、推荐系统;张霄雁,女,1983年生,硕士,工程师,CCF会员,研究方向为空间数据管理、城市计算、数据挖掘;赵路路,女,1995年生,硕士研究生,CCF会员,研究方向为空间数据库管理、Web数据库查询;赵泽祺,女,1994年生,硕士研究生,CCF会员,研究方向为时空众包.
