结合流行度特征和核密度估计的兴趣点推荐算法

发布时间：2018-01-24 03:10

本文关键词： 基于位置社交网络兴趣点推荐流行度特征核密度估计　出处：《小型微型计算机系统》2016年11期 　论文类型：期刊论文

【摘要】：基于位置社交网络中的兴趣点推荐不仅能够帮助用户有效挖掘新地点,带来新体验,还可以帮助兴趣点商家向用户发送广告,提高商业效益.针对兴趣点推荐中存在的精度低问题,提出一种结合流行度特征和核密度估计的兴趣点推荐算法.首先从位置社交网络大量的签到数据中分别提取兴趣点流行度特征和时间连续性特征,然后提出基于连续时间槽的兴趣点流行度评估方法.在此基础上,采用二维高斯核密度估计设计一个兴趣点推荐算法PKDE,最后在三个大型真实社交网络的签到数据集上进行实验对比和分析.实验表明,与已有同类算法相比,本文提出的兴趣点推荐算法能够有效提高推荐精度和召回率.
[Abstract]:The point of interest recommendation in location-based social networks can not only help users to mine new sites and bring new experiences, but also help the merchants of interest points send advertisements to users. Improve the commercial efficiency. The problem of low accuracy in the recommendation of interest points. In this paper, an algorithm of point of interest recommendation based on popularity feature and kernel density estimation is proposed. Firstly, the popular feature of point of interest and the feature of time continuity are extracted from a large number of check-in data of local social network. Then, a continuous time slot based method for evaluating the popularity of interest points is proposed. Based on this, a PKDE algorithm based on two-dimensional Gao Si kernel density estimation is designed. Finally, three large real social network check-in data sets are compared and analyzed. The experimental results show that compared with the existing similar algorithms. The recommendation algorithm proposed in this paper can effectively improve the recommendation accuracy and recall rate.
【作者单位】：燕山大学信息科学与工程学院;燕山大学里仁学院;东北大学秦皇岛分校计算机与通信工程学院;
【基金】：国家自然科学基金项目(61300193,61379116)资助河北省自然科学基金项目(F2015501105,F2015203046)资助
【分类号】：TP391.3
【正文快照】： 1引言基于位置社交网络(Location-Based Social Networks,简称LBSNs)的快速发展,不仅极大丰富了人们的生活体验,还提供社交、位置、时间以及活动内容等多维信息数据.面对LBSNs中海量的信息和生活中数量众多的兴趣点(例如餐厅、旅游景点、酒店等[1]),通过挖掘社交网站用户签到

【相似文献】