Effectively managing offensive content is crucial on social media platforms to encourage positive online interactions. However, addressing offensive contents in code-mixed Dravidian languages faces challenges, as current moderation methods focus on flagging entire comments rather than pinpointing specific offensive segments. This limitation stems from a lack of annotated data and accessible systems designed to identify offensive language sections. To address this, our shared task presents a dataset comprising Kannada-English code-mixed social comments, encompassing offensive comments. This paper outlines the dataset, the utilized algorithms, and the results obtained by systems participating in this shared task.

In recent times, offensive span identification in code-mixed Tamil-English language has seen traction with the release of datasets, shared tasks, and the development of multiple methods. However, the details of various errors shown by these methods are currently unclear. This paper presents a detailed analysis of various errors in state-of-the-art Tamil-English offensive span identification methods. Our study reveals the strengths and weaknesses of the widely used sequence labeling and zero-shot models for offensive span identification. In the due process, we identify data-related errors, improve data annotation and release additional diagnostic data to evaluate models’ quality and stability. 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 emphasize the complexity of various errors and linguistic research challenges.

Maintaining effective control over offensive content is essential on social media platforms to foster constructive online discussions. Yet, when it comes to code-mixed Dravidian languages, the current prevalence of offensive content moderation is restricted to categorizing entire comments, failing to identify specific portions that contribute to the offensiveness. Such limitation is primarily due to the lack of annotated data and open source systems for offensive spans. To alleviate this issue, in this shared task, we offer a collection of Tamil-English code-mixed social comments that include offensive comments. This paper provides an overview of the released dataset, the algorithms employed, and the outcomes achieved by the systems submitted for this task.

Automatic Speech Recognition and its applications are rising in popularity across applications with reasonable inference results. Recent state-of-the-art approaches, often employ significantly large-scale models to show high accuracy for ASR as a whole but often do not consider detailed analysis of performance across low-resource languages applications. In this preliminary work, we propose to revisit ASR in the context of Connected Number Recognition (CNR). More specifically, we (i) present a new dataset HCNR collected to understand various errors of ASR models for CNR, (ii) establish preliminary benchmark and baseline model for CNR, (iii) explore error mitigation strategies and their after-effects on CNR. In the due process, we also compare with end-to-end large scale ASR models for reference, to show its effectiveness.

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.