Research Of Demand Detection And Personality Analysis Based On Micro-blog

With the popularization of social network, micro-blog which contains less than140words to shares messages is playing an important role in everyday life. At the end of December2012, there are about500million registered users on Sina Weibo, and the daily active users reached more than46,000thousands. So it generates enormous data every day. It is a necessary and challengeable task to mine valuable information from these data. This thesis focuses on extracting information from micro-blog in two aspects.First, this thesis researches detecting the demand of the users of micro-blog. The demand information is defined by four tuple in this thesis. Then, the original problem can be regarded as a classify task, whether the demand information could be recognized or not. This thesis extracts features based on words, distance and topics separately under given template. It adopts Support Vector Machine algorithm to implement classify. The results show the method which combine words and distance as features outperforms others.Second, this thesis researches personality on mirco-blog. Traditional personality analysis is principally based on the conventional text. The data of micro-blog is more random and reliable. Based on the model of "Big five personality", the personality analysis could be solved by multivariate classification. This thesis extracts features including words, LIWC and MRC. It has been shown that, LIWC and MRC are more effective on personality analysis.The research mentioned above would help the companies to provide individualized recommendations. Considered the personality of users, the company can push appropriate product or service information when detects the demand of users. The research has practical significance.