摘要
地点网络可从一些独特的视角来刻画城市的空间结构。通过研究城市地点网络的特点及其与传统社交网络的区别,提出了基于地点网络的社区发现算法。该算法综合考虑地点临近性、地点间的连接和用户出行行为的相似性,先进行初始社区的划分,再反复迭代计算各地点隶属于本社区的程度,对隶属度较低的地点进行调整直到收敛,从而发现有意义的城市社区。通过分析社区内部地点的属性和关联,验证了算法的有效性。
The location network can portray the spatial structure of city from some unique perspectives.By studying the characteristics of urban location network and its difference with traditional social network,a community discovery algorithm based on location network was proposed.The algorithm takes into account the proximity of location,the connection between the locations and the similarity of user's travel behavior.Firstly,the initial community is divided.Then,the extent of each site belonging to this community is interatively calculated the places with lower membership degree are adjusted until convergence,so as to find significant urban communities.The validity of the algorithm was verified by analyzing the attributes and correlations of the internal sites.
作者
郑香平
於志勇
温广槟
ZHENG Xiang-ping;YU Zhi-yong;WEN Guang-bin(College of Mathematics and Computer Science,Fuzhou Universit;Fujian Provincial Key Laboratory of Network Computing and Intelligent Information Processing,Fuzhou Universit)
出处
《计算机科学》
CSCD
北大核心
2018年第6期46-50,共5页
Computer Science
基金
国家自然科学基金(61300103)资助
关键词
群智感知
社区发现
地点网络
Crowd sensing
Community discovery
Location network