PKUSE at SemEval-2019 Task 3: Emotion Detection with Emotion-Oriented Neural Attention Network

Proceedings of the 13th International Workshop on Semantic Evaluation (SemEval-2019), pages 287–291 Minneapolis, Minnesota, USA, June 6–7, 2019. ©2019 Association for Computational Linguistics 287 PKUSE at SemEval-2019 Task 3: Emotion Detection with Emotion-Oriented Neural Attention Network Luyao Ma1,2∗, Long Zhang1,2∗, Wei Ye1, Wenhui Hu1 1National Engineering Research Center for Software Engineering, Peking University 2School of Software and Microeconomics, Peking University {maluyao, zhanglong418, wye, huwenhui}@pku.edu.cn Abstract This paper presents the system in SemEval- 2019 Task 3, “EmoContext: Contextual Emo- tion Detection in Text”. We propose a deep learning architecture with bidirectional LSTM networks, augmented with an emotion- oriented attention network that is capable of extracting emotion information from an ut- terance. Experimental results show that our model outperforms its variants and the base- line. Overall, this system has achieved 75.57% for the microaveraged F1 score. 1 Introduction With the rapid development of social media plat- forms like Twitter, a huge number of textual dia- logues has increasingly emerged. It is a challenge for chat bots to generate responses based on user emotions which can avoid inappropriate conversa- tions. Emotion detection in text (Chatterjee et al., 2019) is a research area within Natural Language Processing which is aim to detect the emotion of user expressed in text. Many techniques have been proposed, Wang et al.,Hasan et al.,Liew and Turtle used feature en- gineering to extract features manually. In this area, deep learning-based approches have performed well in recent years. Some methods (W¨ollmer et al., 2010; Metallinou et al., 2012; Poria et al., 2017; Chernykh et al., 2017) used recurrent neu- ral network to model the sequence of utterances for emotion detection. However, those models did not highlight the emotion-related parts. We use at- tention mechanism to locate the parts expressing emotions in the utterance. The Task3 in Semeval-2019 is to detect contex- tual emotions in text. For this task, we propose a deep learning approach which is a combination of ∗These authors have contributed equally to this work. Long Short-Term Memory network and attention mechanism. The rest of the paper is organized as follows: Section 2 provides system overview. Section 3 de- scribes our approach in detail. Our experiment is discussed in Section 4. We conclude our work in Section 5. 2 System Overview 2.1 Text Preprocessing and Word Embedding We use word embeddings as input to the model. Word embeddings are distributed vector presenta- tions of words (Mikolov et al., 2013), capturing their syntactic and semantic information. A good word embedding can get a better classification per- formance. After comparison, we find that the ef- fect of the GloVe (Pennington et al., 2014) is the best, but when we turn the word into a word vec- tor, we find a lot of cases that are out of vocabu- lary(oov). In view of that, we preprocess the data as follows: • The emoji used in the chat can better express human emotions, so we turn them into corre- sponding emotion words and add them to the sentence, which not only solves the …

Content truncated to 500 words