中央研究院 資訊科學研究所

Opinion mining and sentiment analysis is a growing research area. It involves describing, extracting opinions and emotions from text and categorizing them. It is an important area of research because opinions and sentiment information have a great impact on our behaviors. Obviously, it has a great potential and many applications for the future. The research area involves many challenging research topics, such as subjectivity analysis, affect analysis, emotion detection, etc., depending on the objectives. Various thesauri, resources and corpora have been generated for the research community. In this presentation, my major focus will be the techniques of processing and analyzing the Chinese sentiment materials. I will highlight how the unstructured and structured Chinese materials were developed. Chinese opinion mining and sentiment analysis use different techniques depending on whether we want to describe the author or the reader’s perspective, and these will be introduced. Based on them and a move toward Chinese opinion and sentiment detection, a link of the sentiment information and human behaviors can then be constructed. Finally, I will show an application of predicting the user’s likely behavior using sentiment information.