A Topic Study of Microblog Based on Specific Events
This paper describes the study of topics on microblog based on specific events.First,we use a famous topic model-LDA to extract topics from microblog about events.Then,we propose three indexes:Attention Factor(AF),Evolution Factor(EF) to see the performance of microblog topics and Diversity Factor(DF) to calculate the divergence of topics from mieroblog and news reports.Finally,we choose corpuses for four events to study.The experiments show that,on specific events:1) There are more critical topics,while factual topics less,and both of them get close AFs.2) Critical topics last long on microblog and have lower EFs,which means their contents vary little,but factual topics last intermittently and their contents vary greatly.3) To compare with the same events from news reports,critical topics use totally different words,but factual topics use close words.
topic model microblog trend
Zhenyu Zhou Fang Li
Department of Computer Science and Engineering Shanghai Jiao Tong University Shanghai, China, 200240
国际会议
太原
英文
327-330
2013-04-06（万方平台首次上网日期，不代表论文的发表时间）