Semantic Vectors at SemEval-2026 Task 9: Robust Multilingual Polarization Detection via Dual-Encoder Fusion and Expert Ensembling
TL;DR
A Siamese dual-encoder jointly fine-tuning mDeBERTa-v3-base and XLM-RoBERTa-large via 4-bit QLoRA fused with language-specific expert models fused with language-specific expert models through an XGBoost meta-stacker with per-language Platt calibration achieves macro-F1 = 0.797 and accuracy = 0.827 across all 22 languages.
Abstract
We present S EMANTIC V ECTORS , our sys-tem for POLAR@SemEval-2026 Task 9 on multilingual online polarization detection across 22 typologically diverse languages. Polarization is frequently conveyed through implicit rhetorical framing, making cross-lingual detection highly challenging. We address this with a Siamese dual-encoder jointly fine-tuning mDeBERTa-v3-base and XLM-RoBERTa-large via 4-bit QLoRA, fused with language-specific expert models (GBERT, Italian BERT, Swahili BERT) through an XGBoost meta-stacker with per-language Platt calibration. Rather than addressing class imbalance, focal loss functions as a hard-example miner , concentrating gradients on subtly framed instances rather than lexically obvious ones. Combined with per-language threshold optimization, our system achieves macro-F1 = 0.797 and accuracy = 0.827 across all 22 languages.