Community discovery and sentiment mining for Chinese BLOG

Since the emergence of BLOG, it not only represents a new network technology, but also means the beginning of a new life style. How to utilize and mine the BLOG content which contains hidden sentiment and real-time update is a big challenge in the data-mining domain. As most of the existing method for network text´s topic mining is achieved through clustering text´s topic and label which are labeled by hand and often mismatched with the content of the text, the accuracy rating of this method is low. In this paper, we present a BLOG community mining method based on topic features. We define topic features as representation for the major contents of BLOG. Meanwhile, we present a BLOG hidden sentiment continuous evaluation algorithm, called CSETR (Continuous sentiment evaluation with time regulator). CSETR can be used to evaluate the continuous sentiment scores of topic features. Furthermore, it also consider changes in sentiment intensity over time. Lastly, through CSETR we will build up sentiment trend model for Chinese BLOG community.