作者投稿和查稿 主编审稿 专家审稿 编委审稿 远程编辑

计算机工程 ›› 2013, Vol. 39 ›› Issue (4): 184-189. doi: 10.3969/j.issn.1000-3428.2013.04.043

• 人工智能及识别技术 • 上一篇    下一篇

基于用户关系与属性的微博意见领袖挖掘方法

尹衍腾,李学明,蔡孟松   

  1. (重庆大学计算机学院,重庆 400044)
  • 收稿日期:2012-06-04 出版日期:2013-04-15 发布日期:2013-04-12
  • 作者简介:尹衍腾(1987-),男,硕士研究生,主研方向:数据挖掘,电子商务;李学明,教授、博士生导师;蔡孟松,硕士研究生
  • 基金资助:
    国家自然科学基金资助项目(61103114)

Mining Method of Microblog Opinion Leader Based on User Relationship and Attribute

YIN Yan-teng, LI Xue-ming, CAI Meng-song   

  1. (College of Computer Science, Chongqing University, Chongqing 400044, China)
  • Received:2012-06-04 Online:2013-04-15 Published:2013-04-12

摘要: 提出一种结合用户关系与用户属性的挖掘方法。根据微博特征构建微博用户关系网,采用小世界网络理论确定用户的中心性,以此获得基于用户关系的候选意见领袖。通过分析微博用户属性,建立意见领袖影响体系,提出D-means聚类算法,获得基于用户属性的候选意见领袖,结合2种候选意见领袖得到最终意见领袖。实验结果验证该方法在挖掘意见领袖上比现有方法更加准确有效。

关键词: 意见领袖, 微博, 用户关系, 用户属性, 小世界网络, 聚类分析

Abstract: A hybrid data mining approach based on user relationships and attributes is proposed in this paper. The microblog user relationship network is built according to the microblog features. Small-world network theory is used to determine the user's center and generate a candidate set of opinion leaders based on user relationship. Through analysis on microblog user attribute, this paper builds opinion leader affect system, proposes D-means cluster algorithm, and acquires candidate set of opinion leaders. The final opinion leader is got on combination of the two candidate sets. Experimental results demonstrate that the proposed method outperforms more accurate and effective on mining opinion leader than the existing ones.

Key words: opinion leader, microblog, user relationship, user attribute, small-world network, cluster analysis

中图分类号: