SINAI at SemEval-2019 Task 6: Incorporating lexicon knowledge into SVM learning to identify and categorize offensive language in social media

Proceedings of the 13th International Workshop on Semantic Evaluation (SemEval-2019), pages 735–738 Minneapolis, Minnesota, USA, June 6–7, 2019. ©2019 Association for Computational Linguistics 735 SINAI at SemEval-2019 Task 6: Incorporating lexicon knowledge into SVM learning to identify and categorize offensive language in social media Flor Miriam Plaza-del-Arco, M. Dolores Molina-Gonz´alez, M. Teresa Mart´ın-Valdivia, L. Alfonso Ure˜na-L´opez Department of Computer Science, Advanced Studies Center in ICT (CEATIC) Universidad de Ja´en, Campus Las Lagunillas, 23071, Ja´en, Spain {fmplaza, mdmolina, maite, laurena}@ujaen.es Abstract Offensive language has an impact across so- ciety. The use of social media has aggravated this issue among online users, causing suicides in the worst cases. For this reason, it is impor- tant to develop systems capable of identifying and detecting offensive language in text auto- matically. In this paper, we developed a sys- tem to classify offensive tweets as part of our participation in SemEval-2019 Task 6: Offen- sEval. Our main contribution is the integra- tion of lexical features in the classification us- ing the SVM algorithm. 1 Introduction In recent years, with the emergence of social me- dia, the user-generated content on the Web has grown exponentially. This content has the poten- tial to be transmitted quickly, reaching anywhere in the world in a matter of seconds. Due to the ex- change of ideas between users, we find not only positive comments, but also a wide diffusion of aggressive and potentially harmful content. Con- sequently, this type of remarks affects millions of online users. In fact, it has been reported that these incidents have not only created mental and psychological agony to the online users, but have forced people to deactivate their accounts and, in severe cases like cyberbullying, to commit sui- cides (Hinduja and Patchin, 2018). One of the strategies used to deal with aggressive behavior in social media is to monitor or report this type of content. However, this strategy is not entirely fea- sible due to the huge amount of data that is gen- erated daily by users. Therefore, it is necessary to develop systems capable of identifying this type of content on the Web. In order to tackle this problem, firstly it is im- portant to define the toxic language. The toxic lan- guage can be broadly divided into two categories: hate speech and offensive language (Cheng, 2007; Davidson et al., 2017; Gaydhani et al., 2018). Ac- cording to Cambridge Dictionary, hate speech is defined as “public speech that expresses hate or encourages violence towards a person or group based on something such as race, religion, sex, or sexual orientation”. Offensive language is defined as the text which uses hurtful, derogatory or ob- scene terms made by one person to another person. In this paper, we present the system we devel- oped as part of our participation in SemEval-2019 Task 6 OffensEval: Identifying and Categorizing Offensive Language in Social Media) (Zampieri et al., 2019b). In particular, we participated in sub- task A: Offensive language identification. It is a binary classification task and consists of …

Content truncated to 500 words