Exploiting Discourse Relations for Sentiment Analysis

The overall sentiment of a text is critically affected by its discourse structure. By splitting a text into text spans with different discourse relations, we automatically train the weights of different relations in accordance with their importance, and then make use of discourse structure knowledge to improve sentiment classification. In this paper, we utilize explicit connectives to predict discourse relations, and then propose several methods to incorporate discourse relation knowledge to the task of sentiment analysis. All our methods integrating discourse relations perform better than the baseline methods, validating the effectiveness of using discourse relations in Chinese sentiment analysis. We also automatically find out the most influential discourse relations and connectives in sentiment analysis. TITLE AND ABSTRACT IN CHINESE