From prompting to evidence-based translation: A RAG+prompt system for Japanese-Chinese translation and its pedagogical potential
The study developed a Retrieval-Augmented Generation (RAG)+Prompt translation system to improve Japanese-Chinese translation of sentences with noun-modifying clause constructions (NMCCs). The system uses linguistic analysis and retrieved examples to create enhanced prompts for an LLM. Results show that increasing the knowledge base size significantly boosts translation quality, proving this method provides an interpretable and auditable way to enhance complex sentence translation.