基于主题模型的微博语料建模

In this paper we investigate topic modeling methods for text, which has been attracting more attention in the natural language processing field. With the help of topic models and focusing on the characteristics of microblog text, we design a procedure of microblog modeling, including pre-procedure for text, content denoising and processed text modeling. We deal with Twitter's data for denoising and modeling, and extract the most representative topics. After modeling, we analyze the result of modeling and study the topic evolutional process of the extracted topics. We also analyze change of characteristics of topics in time dimension to study the topic evolution of microblog.