A Content-Based Blogger's Sentiment Analysis System

摘要

Nowadays, the information of personal viewpoints booms on the Internet. The sentiment analysis based on the content of blogs is a hot issue in natural language processing. This paper establishes a system of blogger's sentiment analysis which is based on comment target extraction. It is supposed to extract the sentimental information and key events in the blogs as well as the related characters, time and locales. This paper makes an intensive study of its key technology – comment target extraction, and presents a target filtering method based on term-frequency filtration and NN filtering algorithm. It also classifies sentences of sentiment and makes analysis rules for that. According to test results, the system has higher accuracy.