arXiv 2507.169742026-08-04cs.CLcs.AI

利用合成数据结合多语言大语言模型（LLMs）处理农业领域的问答任务

Leveraging Synthetic Data for Question Answering with Multilingual LLMs in the Agricultural Domain

CSIR-Central Scientific Instruments Organisation, India（印度CSIR-中央科学仪器组织）；Academy of Scientific and Innovative Research (AcSIR), Ghaziabad, U.P, India（印度科学与创新研究院（AcSIR））；School of Mathematical and Computational Sciences, Massey University, New Zealand（梅西大学数学与计算科学学院）

Rishemjit Kaur, Arshdeep Singh Bhankhar, Jashanpreet Singh Salh, Sudhir Rajput, Vidhi, Kashish Mahendra, Bhavika Berwal, Ritesh Kumar, Surangika Ranathunga

备注：: 19 pages, 7 tables, Appendix A-Q

阅读 PDF

arXiv 页面 ↗HTML 原文 ↗

AI总结：

本研究针对通用多语言LLMs在农业领域问答中精准性不足的问题，利用印度农业文档生成多语言合成数据集微调LLMs，显著提升了模型的事实性、相关性与农业共识性。

摘要：

让农民能及时获取母语的准确农业信息对农业领域发展至关重要。公开通用型大语言模型（LLMs）通常提供通用农业建议，在本地和多语言场景中缺乏精准性。本研究通过从印度农业专属文档生成多语言（英语、印地语、旁遮普语）合成数据集，并对LLMs进行问答任务微调来解决该局限。在人工构建的数据集上的评估显示，与基线模型相比，微调后的LLMs在事实性、相关性和农业共识方面有显著提升。

英文摘要：

Enabling farmers to access accurate agriculture-related information in their native languages in a timely manner is crucial for the success of the agriculture field. Publicly available general-purpose Large Language Models (LLMs) typically offer generic agriculture advisories, lacking precision in local and multilingual contexts. Our study addresses this limitation by generating multilingual (English, Hindi, Punjabi) synthetic datasets from agriculture-specific documents from India and fine-tuning LLMs for the task of question answering (QA). Evaluation on human-created datasets demonstrates significant improvements in factuality, relevance, and agricultural consensus for the fine-tuned LLMs compared to the baseline counterparts.

机构、总结、中文摘要均由AI分析而来，可能存在错误，请以论文正文为准。