Modeling blog user for Chinese micro-blog sentiment classification with fuzzy SVM

Micro-blog is easily posted and communicated because of its short text length. Meanwhile, its brevity limits the feature expression. This paper presents a blogger modeling method for Chinese micro-blog sentiment classification. The paper would adopt a multidimensional strategy to accomplish the feature extraction task. Compared with traditional feature extraction methods and combined with the interactive property of blog, the blog comment will be considered as the performance characteristic of micro-blog emotion expression. It comes to comments content and the relationship between commentators and micro-bloggers while it extracts the micro blogging text feature. At last, the paper adopts fuzzy SVM model to reduce the sensation of noise points and outliers on classification problem. The results demonstrate the effectiveness of the multidimensional feature extraction method on sentiment classification and the classify effect of fuzzy SVM is better than SVM.