SSN_NLP at SemEval-2019 Task 6: Offensive Language Identification in Social Media using Traditional and Deep Machine Learning Approaches

Proceedings of the 13th International Workshop on Semantic Evaluation (SemEval-2019), pages 739–744 Minneapolis, Minnesota, USA, June 6–7, 2019. ©2019 Association for Computational Linguistics 739 SSN NLP at SemEval-2019 Task 6: Offensive Language Identification in Social Media using Traditional and Deep Machine Learning Approaches D. Thenmozhi, B. Senthil Kumar, Chandrabose Aravindan, S. Srinethe Department of CSE, SSN College of Engineering, India {theni d,senthil,aravindanc}@ssn.edu.in srinethe16108@cse.ssn.edu.in Abstract Offensive language identification (OLI) in user generated text is automatic detection of any profanity, insult, obscenity, racism or vulgar- ity that degrades an individual or a group. It is helpful for hate speech detection, flame de- tection and cyber bullying. Due to immense growth of accessibility to social media, OLI helps to avoid abuse and hurts. In this pa- per, we present deep and traditional machine learning approaches for OLI. In deep learning approach, we have used bi-directional LSTM with different attention mechanisms to build the models and in traditional machine learn- ing, TF-IDF weighting schemes with classi- fiers namely Multinomial Naive Bayes and Support Vector Machines with Stochastic Gra- dient Descent optimizer are used for model building. The approaches are evaluated on the OffensEval@SemEval2019 dataset and our team SSN NLP submitted runs for three tasks of OffensEval shared task. The best runs of SSN NLP obtained the F1 scores as 0.53, 0.48, 0.3 and the accuracies as 0.63, 0.84 and 0.42 for the tasks A, B and C respectively. Our ap- proaches improved the base line F1 scores by 12%, 26% and 14% for Task A, B and C re- spectively. 1 Introduction Offensive language identification (OLI) is a pro- cess of detecting offensive language classes (Razavi et al., 2010) such as slurs, homopho- bia, profanity, extremism, insult, disguise, obscen- ity, racism or vulgarity that hurts or degrades an individual or a group from user-generated text like social media postings. OLI is useful for several applications such as hate speech detec- tion, flame detection, aggression detection and cyber bullying. Recently, several research work have been reported to identify the offensive lan- guages using social media content. Several work- shops such as TA-COS1, TRAC2 (Kumar et al., 2018a), Abusive Language Online3 and GermEval (Wiegand et al., 2018) have been organized re- cently in this research area. In this line, Of- fensEval@SemEval2019 (Zampieri et al., 2019b) shared task focuses on identification and catego- rization of offensive language in social media. It focuses on three subtasks namely offensive lan- guage detection, categorization of offensive lan- guage and offensive language target identification. Sub Task A aims to detect text as offensive (OFF) or not offensive (NOT). Sub Task B aims to cat- egorize the offensive type as targeted text (TIN) or untargeted text (UNT). Sub Task C focuses on identification of target as individual (IND), group (GRP) or others (OTH). Our team SSN NLP par- ticipated in all the three subtasks. 2 Related Work Several research work have been reported since 2010 in this research field of hate speech detection (Kwok and Wang, 2013; Burnap and Williams, 2015; Djuric …

Content truncated to 500 words