ISSN 1000-1239 CN 11-1777/TP

计算机研究与发展 ›› 2015, Vol. 52 ›› Issue (5): 1022-1028.doi: 10.7544/issn1000-1239.2015.20131549

• 信息处理 • 上一篇    下一篇



  1. 1(中国科学院计算技术研究所 北京 100190); 2(国家计算机网络应急技术处理协调中心 北京 100029) (
  • 出版日期: 2015-05-01
  • 基金资助: 

Microblog Bursty Topic Detection Method Based on Momentum Model

He Min1,2, Du Pan1, Zhang Jin1, Liu Yue1, Cheng Xueqi1   

  1. 1(Institute of Computing Technology, Chinese Academy of Sciences, Beijing 100190); 2(National Computer Network Emergency Response Technical Team/Coordination Center of China, Beijing 100029)
  • Online: 2015-05-01

摘要: 针对微博特征空间动态变化、信息噪音大的特点,提出一种基于有意义串动量模型的微博突发话题检测方法.提取时间窗口内微博信息流的有意义串,作为微博信息的动态特征,根据动力学原理对特征进行动量建模,结合特征能量大小、变化趋势以及二阶变化率检测突发特性有意义串,即突发特征,合并突发特征形成突发话题.微博数据实验表明,该方法适用于在线微博突发话题检测,在准确率和召回率上都有明显提升.

关键词: 突发话题, 微博, 突发特征, 有意义串, 动量模型

Abstract: Microblogs reflect the general public's real-time reaction to major events. Finding bursty topics from microblogs is an important task to understand the current events which attract a large number of Internet users. However, the existing methods suitable for news articles aren't adopted directly for microblogs, because microblogs have unique characteristics compared with formal texts, including diversity, dynamic and noise. In this paper, a new detection method for microblog bursty topic is proposed based on momentum model. The meaningful strings are extracted from micorblog posts in the special time window as the microblog dynamic features. The dynamic characteristics of these features are modeled by the principle of momentum. The velocity, accelerated velocity and momentum of the features are defined by the dynamic frequencies at different dimensions. The bursty features are detected with the combination of momentum, variation trend and second order change rate. By merging the detected bursty features with mutual information, the bursty topics are obtained. The experiments are conducted on a real Sina microblog data set containing around 526 thousand posts of 1000 users, and results show that the proposed method improves the precision and recall remarkably compared with the conventional methods. The proposed method could be well applied in online bursty topic detection for microblog information.

Key words: bursty topic, microblog, bursty feature, meaningful string, momentum model