基于链接强度的兴趣相似社区聚类算法  

Link strength-based clustering algorithm of community with similar interest

作　　者：张其文[1] 尚明珍 

机构地区：[1]兰州理工大学计算机与通信学院,甘肃兰州730050

出　　处：《兰州理工大学学报》2017年第2期97-103,共7页Journal of Lanzhou University of Technology

基　　金：国家自然科学基金(61263047);甘肃省自然科学基金(2011GS04147)

摘　　要：传统的社交网络社区划分分为两种,一种是以链接属性进行划分,另一种是以用户自身属性进行划分.近年来出现了融合节点自身属性和链接属性的社区划分算法,但是这些算法只是单纯地将链接属性融为节点自身属性进行划分,忽略了链接属性强弱对节点间兴趣度的反映程度.针对这些问题,对微博中用户的链接属性进行了分类,采用直接链接节点链接关注度和间接链接节点链接关注度的概念,以链接强度为搜索顺序,提出一种基于链接强度的兴趣相似社区划分算法.实验表明,本算法划分的社区内链接度质量较高且用户兴趣相似.The detection of traditional social inter course network is divided into two kinds, one being de- tection with linkage attribute and the other being with attribute of user proper. In recent years, some com- munity detection algorithms have appeared where the attribute of note proper and linkage attribute are fused together. But these algorithms only simply merge the link attribute into the attribute of node prop- er, ignoring the response level of interest degree among the node linkages to the intensity of linkage attrib- ute. Aimed at these problems, the link attributes of micro-blog users are classified and the concept of link attention degree of directly linked nodes and indirectly linked nodes is employed to present a linkage strength-based community detection algorithm where the linkage strength is taken as searching sequence. Experiment shows that by using this algorithm, the inner link strength of community detection will have higher quality and the user＇s interest will be similar.

关 键 词：社区划分 直接链接节点 间接链接节点 链接关注度 兴趣相似度 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]