Reasoning-Guided Claim Normalization for Noisy Multilingual Social Media Posts

TL;DR

This paper presents a multilingual claim normalization method that decomposes social media posts into structured questions, enabling effective cross-lingual misinformation detection with significant improvements over baselines.

Contribution

The study introduces a systematic decomposition approach using W-W-W-H questions and fine-tuning of Qwen3-14B for multilingual claim normalization, achieving robust cross-lingual transfer.

Findings

Achieved METEOR scores from 41.16 (English) to 15.21 (Marathi).

Secured third place on the English leaderboard and fourth on Dutch and Punjabi.

Demonstrated 41.3% relative improvement over baseline methods.

Abstract

We address claim normalization for multilingual misinformation detection - transforming noisy social media posts into clear, verifiable statements across 20 languages. The key contribution demonstrates how systematic decomposition of posts using Who, What, Where, When, Why and How questions enables robust cross-lingual transfer despite training exclusively on English data. Our methodology incorporates finetuning Qwen3-14B using LoRA with the provided dataset after intra-post deduplication, token-level recall filtering for semantic alignment and retrieval-augmented few-shot learning with contextual examples during inference. Our system achieves METEOR scores ranging from 41.16 (English) to 15.21 (Marathi), securing third rank on the English leaderboard and fourth rank for Dutch and Punjabi. The approach shows 41.3% relative improvement in METEOR over baseline configurations and substantial gains over existing methods. Results demonstrate effective cross-lingual generalization for Romance and Germanic languages while maintaining semantic coherence across diverse linguistic structures.