Few-Shot Multilingual Open-Domain QA from 5 Examples

2025-02-27

AI Summary

Low-resource languages in multilingual open-domain question answering (MLODQA) suffer from high annotation costs and severe data scarcity. To address this, we propose FsModQA: (1) a few-shot multilingual data synthesis paradigm leveraging large language models (LLMs), requiring only five exemplars per language to generate high-quality training data; (2) a cross-lingual prompting strategy that enables zero-shot transfer to unseen languages by exploiting English supervision signals; and (3) joint fine-tuning integrating Wikidata-based self-supervised pretraining with multilingual retrieval. Experiments demonstrate that FsModQA significantly outperforms existing baselines under both few-shot and zero-shot settings, achieving state-of-the-art performance on both cross-lingual and monolingual retrieval tasks.

Technology Category

• Machine Learning: Large Multimodal Models (LMMs)
• Natural Language Processing: Language Grounding & Multi-modal NLP
• Computer Vision: Multi-modal Vision

Application Category

• Search and Retrieval-Augmented AI: Multilingual and cross-lingual Web search
• Web Mining and Content Analysis: Mining multimedia, multimodal, multilingual, cross-lingual Web data
• User Modeling, Personalization and Recommendation: Large Language Models (LLM) for user modeling and recommendation

Abstract

Recent approaches to multilingual open-domain question answering (MLODQA) have achieved promising results given abundant language-specific training data. However, the considerable annotation cost limits the application of these methods for underrepresented languages. We introduce a emph{few-shot learning} approach to synthesise large-scale multilingual data from large language models (LLMs). Our method begins with large-scale self-supervised pre-training using WikiData, followed by training on high-quality synthetic multilingual data generated by prompting LLMs with few-shot supervision. The final model, extsc{FsModQA}, significantly outperforms existing few-shot and supervised baselines in MLODQA and cross-lingual and monolingual retrieval. We further show our method can be extended for effective zero-shot adaptation to new languages through a emph{cross-lingual prompting} strategy with only English-supervised data, making it a general and applicable solution for MLODQA tasks without costly large-scale annotation.

Problem

• Multilingual Open-Domain QA
• Few-Shot Learning
• Synthetic Data Generation

Innovation

• Few-shot learning approach
• Synthetic multilingual data generation
• Cross-lingual prompting strategy