DLRG@DravidianLangTech-ACL2022: Abusive Comment Detection in Tamil using Multilingual Transformer Models

Abstract

Online Social Network has let people connect and interact with each other. It does, however, also provide a platform for online abusers to propagate abusive content. The majority of these abusive remarks are written in a multilingual style, which allows them to easily slip past internet inspection. This paper presents a system developed for the Shared Task on Abusive Comment Detection (Misogyny, Misandry, Homophobia, Transphobic, Xenophobia, Coun-terSpeech, Hope Speech) in Tamil Dravidi-anLangTech@ACL 2022 to detect the abusive category of each comment. We approach the task with three methodologies -Machine Learning, Deep Learning and Transformerbased modeling, for two sets of data -Tamil and Tamil+English language dataset. The dataset used in our system can be accessed from the competition on CodaLab. For Machine Learning, eight algorithms were implemented, among which Random Forest gave the best result with Tamil+English dataset, with a weighted average F1-score of 0.78. For Deep Learning, Bi-Directional LSTM gave best result with pre-trained word embeddings. In Transformer-based modeling, we used In-dicBERT and mBERT with fine-tuning, among which mBERT gave the best result for Tamil dataset with a weighted average F1-score of 0.7.

On Finetuning Adapter-Based Transformer Models for Classifying Abusive Social Media Tamil Comments

“…As can be seen in Table 5, adapter-based MuRIL (Large) obtained better accuracy than other proposed and existing models. Even though, XLM-RoBERTa, proposed by [58] gives better performance than models proposed in the current study, MuRIL-large has outperformed this But, as shown in Table 5, the models proposed in [57] have given better performance than proposed To justify the performance of the proposed models, we have compared the parameter e ciency of the transformer based models and the results are discussed in Section 6.3.…”