BITS Pilani at SemEval-2026 Task 9: Structured Supervised Fine-Tuning with DPO Refinement for Polarization Detection

TL;DR

This paper presents a two-stage method combining supervised fine-tuning and DPO refinement on large language models to improve multilingual polarization detection in social media, achieving state-of-the-art results.

Contribution

The novel approach integrates structured supervised fine-tuning with DPO refinement on LLMs for polarization detection, enhancing accuracy over existing baselines.

Findings

Achieved 0.7664 Macro-F1 on English test set.

Post-submission experiments improved Macro-F1 to 0.8162.

Method surpasses the organizer baseline of 0.7802.

Abstract

The POLAR SemEval-2026 Shared Task aims to detect online polarization and focuses on the classification and identification of multilingual, multicultural, and multi-event polarization. Accurate computational detection of online polarization is challenging due to nuanced rhetoric, implicit framing, and the high cost of human-in-the-loop annotation. Building on recent findings that contextual prompting enables large language models to function as strong polarization detectors, we present a two-stage approach for detecting polarization in social media text that combines structured supervised fine tuning with Direct Preference Optimization (DPO) refinement. We fine tune Qwen 2.5-7B-Instruct with LoRA using an interpretable slot-filling template (target, claim type, manifestation checklist, and justification). We then apply DPO with automatically generated preference pairs to reduce costly false negatives. Our submitted system achieves 0.7664 Macro-F1 on the English test set. Post-submission experiments with Mistral-Nemo-Instruct-2407 and LLM-judge-filtered preference pairs further improve to 0.8162 Macro-F1 (not submitted to CodaBench), surpassing the organiser baseline of 0.7802.