YEZE at SemEval-2026 Task 9: Detecting Multilingual, Multicultural and Multievent Online Polarization via Heterogeneous Ensembling

TL;DR

This paper describes a multilingual system for detecting online polarization across 22 languages, utilizing heterogeneous ensemble models and techniques like data augmentation and class weighting to improve performance.

Contribution

The authors introduce a heterogeneous ensemble approach combining XLM-RoBERTa and mDeBERTa models, with techniques to handle severe label imbalance in polarization detection.

Findings

Independent task modeling with class weighting outperforms other methods.

Multi-task learning and translation-based data augmentation contribute to improved accuracy.

The system effectively detects polarized content across multiple languages.

Abstract

This paper presents our system for SemEval-2026 Task 9: Detecting Multilingual, Multicultural and Multievent Online Polarization, which identifies polarized social media content in 22 languages through three subtasks: binary detection, target classification, and manifestation identification. We propose a heterogeneous ensemble of multilingual pretrained models, combining XLM-RoBERTa-large and mDeBERTa-v3-base. We investigate techniques such as multi-task learning, translation-based data augmentation, and class weighting to improve classification performance under severe label imbalance. Our findings indicate that independent task modeling combined with class weighting is more effective.