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 (3): 191-196. doi: 10.3969/j.issn.1000-3428.2013.03.038

• 人工智能及识别技术 • 上一篇    下一篇

基于增量主题模型的微博在线事件分析

马慧芳1,王 博2   

  1. (1. 西北师范大学计算机科学与工程学院,兰州 730070;2. 解放军南京政治学院上海校区军事信息管理系,上海 200433)
  • 收稿日期:2012-04-09 出版日期:2013-03-15 发布日期:2013-03-13
  • 作者简介:马慧芳(1981-),女,副教授、博士,主研方向:数据挖掘,机器学习;王 博,馆员、博士研究生
  • 基金资助:
    国家自然科学基金资助项目(61163039);西北师范大学青年教师科研能力提升计划骨干基金资助项目(NWNU- LKQN-10-1)

Microblog Online Event Analysis Based on Incremental Topic Model

MA Hui-fang 1, WANG Bo 2   

  1. (1. College of Computer Science and Engineering, Northwest Normal University, Lanzhou 730070, China; 2. Department of Military Information Management, Shanghai Branch of PLA Nanjing Institute of Politics, Shanghai 200433, China)
  • Received:2012-04-09 Online:2013-03-15 Published:2013-03-13

摘要: 为更好地利用微博结构化社会网络方面的信息,提出一种基于增量主题模型的微博在线事件分析算法。通过设计增量过程,保留已有的训练信息,采用自适应非对称学习算法融入新微博内容与用户关系。实验结果表明,该算法可在短暂的时间内建模,并有效提高事件分析的性能。

关键词: 用户关系, 话题检测与追踪, 主题模型, 自适应, 增量概率, 增量算法

Abstract: Aiming at the existing event analysis algorithms do not make full use of the structure information on social network of microblogs, this paper proposes a microblog online event analysis algorithm based on incremental topic model. This algorithm designs a reasonable incremental process to preserve the existing training information, and gives an adaptive asymmetric learning mechanism to integrate the content and user relationship of new microblogs. Experimental results show that this algorithm leads to more balanced and comprehensive improvement for online event detection in near real-time scenarios.

Key words: user relationship, Topic Detection and Tracking(TDT), topic model, adaptive, incremental probability, incremental algorithm

中图分类号: