Abstract: In recent years,since the rise of personal-media caused a huge public opinion crisis,how to discover hot topics from the fragmentation of the mass microblogging information and extract useful information has become a major challenge.In this background,we proposed an extract summarization method based on improved Lex-PageRank algorithm.In this program,we make a simple clustering and use these clustering results as experimental data.With due consideration for the time characteristics of microblog influence cycle and weight attribute,the improved Lex-PageRank algorithm is combined with MMR algorithm to get many texts from clustering results to generate a summary.The experiment based on Sina Weibo indicates that our method can extract critical information effectively from mass texts.

Key words: Microblog,Time characteristics,Weight attribute,Lex-PageRank algorithm