计算机应用 ›› 2010, Vol. 30 ›› Issue (05): 1268-1272.

• 数据挖掘与人工智能 • 上一篇    下一篇

基于节点动态属性相似性的社会网络社区推荐算法

陈琼1,李辉辉2,肖南峰3   

  1. 1. 华南理工大学 计算机科学与工程学院
    2.
    3. 广州华南理工大学计算机科学与工程学院
  • 收稿日期:2009-11-04 修回日期:2009-11-29 发布日期:2010-05-04 出版日期:2010-05-01
  • 通讯作者: 陈琼
  • 基金资助:
    国家自然科学基金与中国民用航空总局联合基金资助项目;广东省自然科学基金重点项目

Community recommendation algorithm based on dynamic attributes similarity of nodes in social networks

  • Received:2009-11-04 Revised:2009-11-29 Online:2010-05-04 Published:2010-05-01
  • Contact: CHEN Qiong

摘要: 社区推荐帮助用户寻找感兴趣的社群,是社会网络分析的重要内容。根据社会网络的动态变化特性,通过分析网络的动态演变过程、网络个体的行为特征及个体间联系的变化,研究动态社区及其个体的动态特性的形式化描述方法,提出了节点(个体)间的动态属性相似度计算方法和基于节点(个体)间的动态属性相似度计算的社区推荐算法,可以克服通过个体的直接联系进行社区推荐的局限性。实验结果表明,应用本算法进行社区推荐的准确率有较大提高,能有效应用于动态社会网络的社区推荐。

关键词: 动态社区, 个体活跃度, 群体结构, 节点动态属性, 社区推荐

Abstract: Community recommendation helps users find their interested communities and is becoming an important task in social networks analysis. According to the dynamic property of social networks,the evolvement of social networks,the behavior properties of individuals and the link relationships of nodes,the formal representation methods for the dynamic community and its nodes' dynamic attributes were studied. The computation method for dynamic attributes similarity of nodes and a community recommendation algorithm based on dynamic attributes similarity of nodes were proposed. The proposed algorithm can overcome the limitation in the other algorithms, recommending community to an individual only through its direct linking individuals. The experimental results show that the recommendation precision is improved and the method is effective to the community recommendation in dynamic social networks.

Key words: dynamic community, individual activity, group structure, dynamic attributes of node, community recommendation