The Titans at SemEval-2019 Task 5: Detection of hate speech against immigrants and women in Twitter

Proceedings of the 13th International Workshop on Semantic Evaluation (SemEval-2019), pages 494–497 Minneapolis, Minnesota, USA, June 6–7, 2019. ©2019 Association for Computational Linguistics 494 The Titans at SemEval-2019 Task 5: Detection of hate speech against immigrants and women in Twitter Avishek Garain Computer Science and Engineering Jadavpur University, Kolkata avishekgarain@gmail.com Arpan Basu Computer Science and Engineering Jadavpur University, Kolkata arpan0123@gmail.com Abstract This system paper is a description of the sys- tem submitted to “SemEval-2019 Task 5” Task B for the English language, where we had to primarily detect hate speech and then detect aggressive behaviour and its target audience in Twitter. There were two specific target audi- ences, immigrants and women. The language of the tweets was English. We were required to first detect whether a tweet is containing hate speech. Thereafter we were required to find whether the tweet was showing aggressive be- haviour, and then we had to find whether the targeted audience was an individual or a group of people. 1 Introduction Hate speech attacks a person or a group on the ba- sis of attributes such as race, religion, ethnic ori- gin, national origin, sex, disability, sexual orienta- tion or gender identity. In the same time, flames (such as rants, taunts, and squalid phrases) are of- fensive/abusive phrases which might attack or of- fend the users for a variety of reasons. This is very pertinent due to rise of text messaging through the Internet or cellular phones, which has become a major medium of personal and commercial com- munication. Aggression is overt, often harmful, social inter- action with the intention of inflicting damage or other unpleasantness upon another individual. It may occur either in retaliation or without provo- cation. In humans, frustration due to blocked goals can cause aggression. Human aggression can be classified into direct and indirect aggres- sion; whilst the former is characterized by physi- cal or verbal behavior intended to cause harm to someone, the latter is characterized by behavior intended to harm the social relations of an indi- vidual or group. Hate speech and offensive language are perva- sive in social media. Online communities, social media platforms, and technology companies have been researching heavily in ways to cope with this phenomena to prevent abusive behavior in social media. This is due to text messaging through the Internet or cellular phones, which has become a major medium of personal and commercial com- munication. One of the most effective strategies for tack- ling this problem is to use computational meth- ods to identify hate speech and aggression in user- generated content (e.g. posts, comments, tweets etc.). This topic has attracted significant attention in recent years of various Natural Language ana- lysts. The SemEval 2019 task 5 (Basile et al., 2019) was a classification task where we were required to classify a tweet as containing hate speech or other- wise. However, there were some additional chal- lenges presented, which involved automatic detec- tion of aggression, and classification the target au- dience as an individual or …

Content truncated to 500 words