Public Opinion Analysis of Microblog Content

In this paper, a public opinion analysis system is built up. It consists of a crawler used to retrieve online microblog content and a text classifier for distinguishing sentimental content. This system is used to identify public opinions towards certain topics. Microblogs are divided into three categories based on their emotional tendency, namely "positive", "negative" and "objective", and then the microblogs are classified according to this categories. The classifier is constructed applying Support Vector Machine algorithm and the precision of classification exceeds 90%.