Zero-shot Cross-lingual NER via Mitigating Language Difference: An Entity-aligned Translation Perspective

2025-09-01

AI Summary

Existing zero-shot cross-lingual named entity recognition (ZCL-NER) methods suffer significant performance degradation on non-Latin-script languages (e.g., Chinese, Japanese), primarily due to deep linguistic structural disparities that impede effective knowledge transfer. To address this, we propose an LLM-based entity-aligned translation framework. Our method introduces: (1) a bidirectional translation mechanism—forward (source→target) and back-translation—that explicitly models cross-lingual entity alignment and semantic consistency; and (2) fine-tuning of multilingual LLMs on Wikipedia-derived parallel entity data to enhance cross-script and cross-structural entity mapping capabilities. Experiments across multiple low-resource non-Latin languages demonstrate substantial improvements in zero-shot transfer performance. The approach effectively mitigates negative transfer induced by typological divergence, offering a novel paradigm for generalizing ZCL-NER across morphologically and structurally dissimilar languages.

Technology Category

• Natural Language Processing: Machine Translation, Multilinguality, Cross-Lingual NLP
• Machine Learning: Large Multimodal Models (LMMs)
• Computer Vision: Language and Vision

Application Category

• Search and Retrieval-Augmented AI: Multilingual and cross-lingual Web search
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs

Abstract

Cross-lingual Named Entity Recognition (CL-NER) aims to transfer knowledge from high-resource languages to low-resource languages. However, existing zero-shot CL-NER (ZCL-NER) approaches primarily focus on Latin script language (LSL), where shared linguistic features facilitate effective knowledge transfer. In contrast, for non-Latin script language (NSL), such as Chinese and Japanese, performance often degrades due to deep structural differences. To address these challenges, we propose an entity-aligned translation (EAT) approach. Leveraging large language models (LLMs), EAT employs a dual-translation strategy to align entities between NSL and English. In addition, we fine-tune LLMs using multilingual Wikipedia data to enhance the entity alignment from source to target languages.

Problem

• Addressing performance degradation in zero-shot cross-lingual NER for non-Latin script languages
• Mitigating structural language differences between Latin and non-Latin script languages
• Improving entity alignment through translation strategies and LLM fine-tuning

Innovation

• Entity-aligned translation using LLMs
• Dual-translation strategy for alignment
• Fine-tuning with multilingual Wikipedia data