Word-Level Emotion Embedding Based on Semi-Supervised Learning for Emotional Classification in Dialogue

Young-Jun Lee, Chan Yong Park, Ho-Jin Choi

Abstract: Emotion classification has been remarkable studies in recent years. However, most of works do not consider the context information such as a flow of emotions. In this paper, we propose the emotion classification in dialogue based on the semi-supervised word-level emotion embedding. For the word-level emotion embedding, we use the NRC Emotion Lexicon which is a list of English words and their associations with eight basic emotions. By adding word-level emotion vectors, we obtain an utterance-level emotion vector. We train a single layer LSTM-based classification network in dialogue. Also, we will evaluate our model on the EmotionLines which is dataset with emotions labeling on all utterances in each dialogue. The experiment plan is described in this paper.