cs.CL提交于 2026-10-22已译

从提示到证据驱动翻译：面向日汉翻译的 RAG+Prompt 系统及其教学潜力

From prompting to evidence-based translation: A RAG+prompt system for Japanese-Chinese translation and its pedagogical potential

Wenshi Gu

中文摘要

大语言模型在高资源语言对翻译中表现良好，但处理含名词性修饰从句(noun-modifying clause constructions, NMCCs)的日语句子时可靠性较低。研究评估了一种检索增强生成(RAG)+Prompt 翻译系统，在不修改基础模型的情况下整合语言分析、基于嵌入的检索、提示构建与 LLM 生成。分析模块输出 A1（内嵌型或外嵌型 NMCC）和 A2（词汇选择、NMCC 处理、语序及文体/语域的风险预测），并按 L2 距离检索 top-k = 5 个相似日汉实例，将实例及 A1/A2 写入增强提示。使用 GPT-4o 和 66 句测试集，比较六种知识库规模（0/100/200/500/1,000/2,000），仅采用句级宏平均 BLEU（1-4-gram，含 brevity penalty；区分大小写；中文按字符计算）。平均 BLEU 从关闭 RAG 时的 24.28 升至知识库规模 2,000 时的 29.96，增加 5.68（23.4%）；各规模下总体呈上升趋势，知识库越大得分越高。该系统以可解释、可审计的方式改善 NMCC 日汉翻译，但局限包括仅测试一个基础模型和一种指标，并依赖已发表文本及商业 API。

关键要点

01指出 NMCC 日语句子的汉译可靠性较低这一语言对差距
02以语言分析、嵌入检索和增强提示整合 GPT-4o
0366 句测试覆盖六种知识库规模及单一 BLEU 指标
04知识库增至 2,000 时平均 BLEU 提高 5.68
05评估受单一模型、指标、文本来源和商业 API 限制

解读

尚无解读。

原始英文摘要

arXiv:2605.03387v2 Announce Type: replace Abstract: Large language models perform well on high-resource pairs but are less reliable for Japanese-Chinese sentences containing noun-modifying clause constructions (NMCCs). This study evaluates a retrieval-augmented generation RAG+Prompt translation system that integrates linguistic analysis, embedding-based retrieval, prompt construction, and LLM generation without modifying the base model. The analysis module outputs A1 (inner vs. outer NMCC) and A2 (risk predictions: lexical choice/NMCC handling/word order/style/register); top-k = 5 similar Ja-Zh examples (L2 distance) and A1/A2 are inserted into an enhanced prompt. Using GPT-4o and a 66-sentence test set, we compare six knowledge-base sizes (0/100/200/500/1,000/2,000). Macro-averaged sentence-level BLEU (1-4-gram with brevity penalty; cased; Chinese at the character level) is the sole metric. Mean BLEU increases from 24.28 at 0 (RAG disabled) to 29.96 at 2,000 (+5.68; +23.4%). The upward trend holds across sizes, with larger knowledge bases yielding higher scores. We conclude that the RAG+Prompt translation system improves Ja-Zh translation of sentences containing NMCCs in an interpretable and auditable manner. Limitations include one base model, one metric, and reliance on published texts and commercial APIs; future work will broaden genres, language pairs, and evaluation metrics.