Enhancing Low-Resource Minority Language Translation with LLMs and Retrieval-Augmented Generation for Cultural Nuances
Fuente:
arXiv
Gespeichert in:
| Hauptverfasser: | , , , |
|---|---|
| Format: | Preprint |
| Veröffentlicht: |
2025
|
| Schlagworte: | |
| Online-Zugang: | |
| Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
| _version_ | 1866912379316469760 |
|---|---|
| author | Chang, Chen-Chi Li, Chong-Fu Lee, Chu-Hsuan Lee, Hung-Shin |
| author_facet | Chang, Chen-Chi Li, Chong-Fu Lee, Chu-Hsuan Lee, Hung-Shin |
| contents | This study investigates the challenges of translating low-resource languages by integrating Large Language Models (LLMs) with Retrieval-Augmented Generation (RAG). Various model configurations were tested on Hakka translations, with BLEU scores ranging from 12% (dictionary-only) to 31% (RAG with Gemini 2.0). The best-performing model (Model 4) combined retrieval and advanced language modeling, improving lexical coverage, particularly for specialized or culturally nuanced terms, and enhancing grammatical coherence. A two-stage method (Model 3) using dictionary outputs refined by Gemini 2.0 achieved a BLEU score of 26%, highlighting iterative correction's value and the challenges of domain-specific expressions. Static dictionary-based approaches struggled with context-sensitive content, demonstrating the limitations of relying solely on predefined resources. These results emphasize the need for curated resources, domain knowledge, and ethical collaboration with local communities, offering a framework that improves translation accuracy and fluency while supporting cultural preservation. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2505_10829 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | Enhancing Low-Resource Minority Language Translation with LLMs and Retrieval-Augmented Generation for Cultural Nuances Chang, Chen-Chi Li, Chong-Fu Lee, Chu-Hsuan Lee, Hung-Shin Computation and Language Artificial Intelligence This study investigates the challenges of translating low-resource languages by integrating Large Language Models (LLMs) with Retrieval-Augmented Generation (RAG). Various model configurations were tested on Hakka translations, with BLEU scores ranging from 12% (dictionary-only) to 31% (RAG with Gemini 2.0). The best-performing model (Model 4) combined retrieval and advanced language modeling, improving lexical coverage, particularly for specialized or culturally nuanced terms, and enhancing grammatical coherence. A two-stage method (Model 3) using dictionary outputs refined by Gemini 2.0 achieved a BLEU score of 26%, highlighting iterative correction's value and the challenges of domain-specific expressions. Static dictionary-based approaches struggled with context-sensitive content, demonstrating the limitations of relying solely on predefined resources. These results emphasize the need for curated resources, domain knowledge, and ethical collaboration with local communities, offering a framework that improves translation accuracy and fluency while supporting cultural preservation. |
| title | Enhancing Low-Resource Minority Language Translation with LLMs and Retrieval-Augmented Generation for Cultural Nuances |
| topic | Computation and Language Artificial Intelligence |
| url | https://arxiv.org/abs/2505.10829 |