Analysis and Generation of Emotion in Texts

This paper explores the task of automatic emotion analysis and generation in texts. We present preliminary results for the task of classifying texts by classes of emotions. Then, we present detailed experiments in classifying texts by classes of mood. We propose a novel approach that uses the hierarchy of possible moods in order to achieve better results than a standard flat classification. We also show that using sentiment orientation features improves the performance of classification. At the end, the possibility of generating texts that express specific emotions is discussed.