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

RAG+提示系统通过语言分析提高日语-中文翻译准确性

作者 PulseAugur 编辑部 · [2 个来源] · 2026-05-05 05:48

研究人员开发了一种结合了检索增强生成（RAG）技术和提示工程的系统，以提高日语-中文机器翻译的准确性，特别是在处理名词修饰从句结构（NMCCs）的句子时。该系统集成了语言分析、基于嵌入的检索和提示工程，以增强GPT-4o等大型语言模型的输出。通过不同知识库大小的测试显示，BLEU分数显著提高，知识库越大效果越好，证明了一种可解释且可审计的翻译改进方法。 AI

影响这种RAG+提示系统为提高大型语言模型在复杂语言结构上的翻译质量提供了一种更具可解释性和可审计性的方法。

排序理由该集群包含一篇详细介绍一种新的用于机器翻译的RAG+提示系统的学术论文。

在 arXiv cs.CL 阅读 →

AI 生成摘要 · Google Gemini · 来自 2 个来源。我们如何撰写摘要 →

报道来源 [2]

arXiv cs.CL TIER_1 English(EN) · Wenshi Gu · 2026-05-06 04:00

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

arXiv:2605.03387v1 Announce Type: new 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 …
arXiv cs.CL TIER_1 English(EN) · Wenshi Gu · 2026-05-05 05:48

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

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 an…

报道来源 [2]

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

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

相关实体

相关话题