TECHSSN at SemEval-2019 Task 6: Identifying and Categorizing Offensive Language in Tweets using Deep Neural Networks

Proceedings of the 13th International Workshop on Semantic Evaluation (SemEval-2019), pages 753–758 Minneapolis, Minnesota, USA, June 6–7, 2019. ©2019 Association for Computational Linguistics 753 TECHSSN at SemEval-2019 Task 6: Identifying and Categorizing Offensive Language in Tweets using Deep Neural Networks Logesh Balasubramanian, Harshini Sathish Kumar, Geetika Bandlamudi, Dyaneswaran Sivasankaran, Rajalakshmi Sivanaiah, Angel Deborah Suseelan, Sakaya Milton Rajendram, Mirnalinee Thanka Nadar Thanagathai Department of Computer Science and Engineering SSN College of Engineering Chennai 603 110, Tamil Nadu, India {logesh16054, harshini16040}@cse.ssn.edu.in, {geetika16032, dyaneswaran16028}@cse.ssn.edu.in, {rajalakshmis, angeldeborahs}@ssn.edu.in {miltonrs, mirnalineett}@ssn.edu.in Abstract Task 6 of SemEval 2019 involves identify- ing and categorizing offensive language in social media. The systems developed by TECHSSN team uses multi-level classification techniques. We have developed two systems. In the first system, the first level of classifica- tion is done by a multi-branch 2D CNN classi- fier with Google’s pre-trained Word2Vec em- bedding and the second level of classification by string matching technique supported by of- fensive and bad words dictionary. The second system uses a multi-branch 1D CNN classi- fier with Glove pre-trained embedding layer for the first level of classification and string matching for the second level of classification. Input data with a probability of less than 0.70 in the first level are passed on to the second level. The misclassified examples are classi- fied correctly in the second level. 1 Introduction The growth of social media networks in recent days has been phenomenal and Twitter is no ex- ception. However, this rapid growth of social me- dia also poses a serious challenge of maintaining ethics in social media because of the degradation of moral values in society. Offensive micro tweets are generated on a daily basis targeting a partic- ular person, organization, race, caste, community, religion, gender and so forth. For this reason, our task for the SemEval2019 (Zampieri et al., 2019b) mainly focuses on the detection of offensive lan- guage in tweets and classify them. In Subtask-A, we tried to classify the tweets into two classes, namely offensive and non-offensive. The offensive tweets in the Subtask-A are then categorized in Subtask-B into targeted and untar- geted tweets, where targeted tweets are aimed at a specific person, organization, religion or politi- cal parties. Further, Subtask-C deals with the fine- grained classification of offensive tweets into three classes viz person, organization and others. The training dataset provided by the organiz- ers contains 13240 tweets. The given dataset is used as the preliminary dataset to train our model. In addition, Impermium dataset from Kaggle and TRAC dataset are added to improve the accuracy of the model. We have also used a dictionary of offensive words in the second level of classifica- tion. Manually classifying the tweets is ambigu- ous and highly subjective, and is one of the biggest challenges. The mix of colloquial slang in tweets, veiled references, missing data, usage of symbols and emojis are further hurdles that lowered the prediction accuracy (Founta et al., 2018). 2 Related Work Many researchers in the field of Artificial Intelli- gence and Natural …

Content truncated to 500 words