Toxic span identification in Tamil is a shared task that focuses on identifying harmful content, contributing to offensiveness. In this work, we have built a model that can efficiently identify the span of text contributing to offensive content. We have used various transformer-based models to develop the system, out of which the fine-tuned MuRIL model was able to achieve the best overall character F1-score of 0.4489.

This paper investigates the effectiveness of sentence-level transformers for zero-shot offensive span identification on a code-mixed Tamil dataset. More specifically, we evaluate rationale extraction methods of Local Interpretable Model Agnostic Explanations (LIME) (CITATION) and Integrated Gradients (IG) (CITATION) for adapting transformer based offensive language classification models for zero-shot offensive span identification. To this end, we find that LIME and IG show baseline F₁ of 26.35% and 44.83%, respectively. Besides, we study the effect of data set size and training process on the overall accuracy of span identification. As a result, we find both LIME and IG to show significant improvement with Masked Data Augmentation and Multilabel Training, with F₁ of 50.23% and 47.38% respectively. Disclaimer : This paper contains examples that may be considered profane, vulgar, or offensive. The examples do not represent the views of the authors or their employers/graduate schools towards any person(s), group(s), practice(s), or entity/entities. Instead they are used to emphasize only the linguistic research challenges.

Offensive content moderation is vital in social media platforms to support healthy online discussions. However, their prevalence in code-mixed Dravidian languages is limited to classifying whole comments without identifying part of it contributing to offensiveness. Such limitation is primarily due to the lack of annotated data for offensive spans. Accordingly, in this shared task, we provide Tamil-English code-mixed social comments with offensive spans. This paper outlines the dataset so released, methods, and results of the submitted systems.

This paper presents the Dravidian Offensive Span Identification Dataset (DOSA) for under-resourced Tamil-English and Kannada-English code-mixed text. The dataset addresses the lack of code-mixed datasets with annotated offensive spans by extending annotations of existing code-mixed offensive language identification datasets. It provides span annotations for Tamil-English and Kannada-English code-mixed comments posted by users on YouTube social media. Overall the dataset consists of 4786 Tamil-English comments with 6202 annotated spans and 1097 Kannada-English comments with 1641 annotated spans, each annotated by two different annotators. We further present some of our baseline experimental results on the developed dataset, thereby eliciting research in under-resourced languages, leading to an essential step towards semi-automated content moderation in Dravidian languages. The dataset is available in https://github.com/teamdl-mlsg/DOSA

In this paper, we present our participation in SemEval-2020 Task-12 Subtask-A (English Language) which focuses on offensive language identification from noisy labels. To this end, we developed a hybrid system with the BERT classifier trained with tweets selected using Statistical Sampling Algorithm (SA) and Post-Processed (PP) using an offensive wordlist. Our developed system achieved 34th position with Macro-averaged F1-score (Macro-F1) of 0.90913 over both offensive and non-offensive classes. We further show comprehensive results and error analysis to assist future research in offensive language identification with noisy labels.