Predicting Reader's Emotion on Chinese Web News Articles

Currently, more and more information are spreading on the web. These large amounts of information might influence web users' emotion quite a lot, for example, make people angry. Thus, it is important to analyze web textual content from the aspect of emotion. Although much former researches have been done, most of them focus on the emotion of authors but not readers. In this paper, we propose a novel method to predict readers' emotion based on content analysis. We develop an emotion dictionary with a selected weighting coefficient to build text vectors in Vector Space Model, and train Support Vector Machine and Naive Bayesian model for prediction. The experimental results indicate that our approach performs much better on precision, recall and F-value.