计算机工程与应用 ›› 2015, Vol. 51 ›› Issue (7): 118-123.

• 数据库、数据挖掘、机器学习 • 上一篇    下一篇

融合位置和社交属性的热点轨迹聚类算法

袁怀旺1,李积丰2,徐  彪1,霍  欢1   

  1. 1.上海理工大学 光电信息与计算机工程学院,上海 200093
    2.菲沙·河谷大学 计算机信息系统系,加拿大 V2S 7M8
  • 出版日期:2015-04-01 发布日期:2015-03-31

Hot trajectory clustering algorithm based on location and social network characteristics

YUAN Huaiwang1, LI Jifeng2, XU Biao1, HUO Huan1   

  1. 1.School of Optical-Electrical and Computer Engineering, University of Shanghai for Science and Technology, Shanghai 200093, China
    2.Department of Computer Information System, University of the Fraser Valley, BC V2S 7M8, Canada
  • Online:2015-04-01 Published:2015-03-31

摘要: 提出一种新的热点轨迹聚类算法(LSHT),结合社交网络中的位置属性和社交特征,利用基于位置的社交网络(LBSN)用户中的好友关系特征,引入用户亲密度和用户活跃度,对用户之间的联系进行分析;提出对用户轨迹进行评分的评分函数,计算用户轨迹的影响力。结合用户亲密度和轨迹评分值,挖掘用户好友中的热点轨迹。该算法能够将热门轨迹中的不真实轨迹进行过滤。实验表明,该算法能更快更准确地挖掘出用户好友中的热点轨迹。

关键词: 社交网络, 轨迹, 用户亲密度, 位置信息, 聚类

Abstract: This paper presents a new clustering algorithm for Location-based Social-network Hot Trajectory(LSHT), which combines social network characteristics into the location-based hot trajectory discovery. The algorithm introduces friendship intimacy and friendship activeness to analyze the relationship among users. Meanwhile, the algorithm introduces a ranking function for hot trajectory based on calculating the influence of the users’ trajectories. Combining friendship intimacy and ranking function, the algorithm digs out the hot trajectory from friends’ trajectories by filtering out the fake trajectories. The experiments show the LSHT can accurately and effectively discover the hot trajectory in a location-based social network.

Key words: social network, trajectory, friendship intimacy, location information, clustering