YNU_DYX at SemEval-2019 Task 5: A Stacked BiGRU Model Based on Capsule Network in Detection of Hate

Proceedings of the 13th International Workshop on Semantic Evaluation (SemEval-2019), pages 535–539 Minneapolis, Minnesota, USA, June 6–7, 2019. ©2019 Association for Computational Linguistics 535 YNU DYX at SemEval-2019 Task 5: A Stacked BiGRU Model Based on Capsule Network in Detection of Hate Yunxia Ding, Xiaobing Zhou∗, Xuejie Zhang School of Information Science and Engineering Yunnan University, Yunnan, P.R. China yxding01@163.com, zhouxb@ynu.edu.cn, xjzhang@ynu.edu.cn Abstract This paper describes our system designed for SemEval 2019 Task 5 “Shared Task on Mul- tilingual Detection of Hate”. We only par- ticipate in subtask-A in English. To address this task, we present a stacked BiGRU mod- el based on a capsule network system. In or- der to convert the tweets into corresponding vector representations and input them into the neural network, we use the fastText tools to get word representations. Then, the sentence representation is enriched by stacked Bidirec- tional Gated Recurrent Units (BiGRUs) and used as the input of capsule network. Our sys- tem achieves an average F1-score of 0.546 and ranks 3rd in the subtask-A in English. 1 Introduction Hate speech is an offensive language, a statemen- t that a person or group attacks another person or group based on characteristics such as gen- der, race, religion, disability, or sexual orientation. Nockleby (Nockleby, 2000) defines hate speech as “any communication that disparages a person or a group on the basis of some characteristic such as race, color, ethnicity, gender, sexual orientation, nationality, religion, or other characteristic.” Giv- en the huge amount of user-generated content on the Web, and in particular on social media, the problem of detecting, and therefore possibly limit the Hate Speech diffusion, is becoming fundamen- tal, for instance for fighting against misogyny and xenophobia (Basile et al., 2019). Microblog today has become a very popular communication tool among Internet users. Mil- lions of users share opinions on different aspects of life everyday. And Twitter1 is a social platform that is very popular all over the word and million- s of people share their experiences, moods, atti- ∗Corresponding author 1http://twitter.com tudes toward life and discuss current issues (Pak and Paroubek, 2010). Many of the content is re- lated to people’s feelings, so many people begin to conduct emotional analysis and research on tweet- s. SemEval 2019 Task 5 is to detect hate speech on tweets. Task A is a binary classification task that predicts whether English or Spanish tweets for specific goals (women or immigrants) are hateful or not hateful (Basile et al., 2019). There are many studies that currently use tweets as a corpus for natural language processing (NLP). Text classifi- cation using traditional machine learning methods mainly includes Support Vector Machines (SVM- s) (Gunn et al., 1998), Naive Bayes (McCallum et al., 1998) and Random Forests (Cutler et al., 2007), etc. In recent years, the use of deep neu- ral networks for NLP has become mainstream, such as Convolutional Neural Networks (CNNs) for sentence classification (Kim, 2014) and Re- current Neural Networks (RNNs) (Graves et al., 2013). This task aims …

Content truncated to 500 words