An Opinion Mining Technique For Chinese Blogs

Nowadays blogs have gained popularity. They are of interest to researches in Information Retrieval. Many authors write opinions about things in their blogs. Opinion analysis in Chinese is a challenging task. In this paper, we discuss blog opinion analysis issues. We use blog documents provided by PIXNET (Taiwanese Internet service provider company) to do experiments. The proposed technique applies institute of computing technology, Chinese lexical analysis system (ICTCLAS) to segment the sentences, then utilizes the NTUS dictionary (National Taiwan university Sentiment Dictionary) to find opinion terms in every blog. After that, we use a support vector machine to define the opinions in the blogs. It detects both the positive and negative opinions in the selected blogs. The experiments demonstrate the promising results.