# NJUST-KMG at TRAC-2024 Tasks 1 and 2: Offline Harm Potential Identification

arXiv: https://arxiv.org/abs/2403.19713 · HTML: https://mlsanity.com/paper/2403.19713/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Builds on 8 of its 22 references that are in the map.
- First posted on arXiv 26 Mar 2024.

## Builds on (most cited first, 8 of 8)

- [BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding](https://mlsanity.com/paper/1810.04805/) (2018)
- [A Simple Framework for Contrastive Learning of Visual Representations](https://mlsanity.com/paper/2002.05709/) (2020)
- [Unsupervised Cross-lingual Representation Learning at Scale](https://mlsanity.com/paper/1911.02116/) (2019)
- [MuRIL: Multilingual Representations for Indian Languages](https://mlsanity.com/paper/2103.10730/) (2021)
- [Semi-Supervised Multi-Modal Multi-Instance Multi-Label Deep Network with Optimal Transport](https://mlsanity.com/paper/2104.08489/) (2021)
- [Hate Speech and Offensive Content Detection in Indo-Aryan Languages: A Battle of LSTM and Transformers](https://mlsanity.com/paper/2312.05671/) (2023)
- [LeOCLR: Leveraging Original Images for Contrastive Learning of Visual Representations](https://mlsanity.com/paper/2403.06813/) (2024)
- [HarmPot: An Annotation Framework for Evaluating Offline Harm Potential of Social Media Text](https://mlsanity.com/paper/2403.11108/) (2024)

## Abstract (arXiv)

This report provide a detailed description of the method that we proposed in the TRAC-2024 Offline Harm Potential dentification which encloses two sub-tasks. The investigation utilized a rich dataset comprised of social media comments in several Indian languages, annotated with precision by expert judges to capture the nuanced implications for offline context harm. The objective assigned to the participants was to design algorithms capable of accurately assessing the likelihood of harm in given situations and identifying the most likely target(s) of offline harm. Our approach ranked second in two separate tracks, with F1 values of 0.73 and 0.96 respectively. Our method principally involved selecting pretrained models for finetuning, incorporating contrastive learning techniques, and culminating in an ensemble approach for the test set.