Sakaji-Lab/LATGNJ

JAgriN（长崎县日本农业数据集）是一个基于长崎县公开的农业指导方针创建的、特定地区的、结构化的日语数据集。它包含了自然语言中的程序性和情境性知识，可用于研究和开发大型语言模型（LLM）进行农业问答、分类和生成任务。该数据集包含1152个步骤，涵盖34个农产品项目，并以JSON和CSV格式提供。它适用于自然语言处理应用，也适用于其他机器学习任务，如时间推理、程序建模和数据到文本的生成。

JAgriN is a region-specific, structured Japanese-language dataset derived from publicly available agricultural guidelines issued by Nagasaki Prefecture, Japan. It captures procedural and contextual knowledge in natural language, enabling research and development of Large Language Models (LLMs) for agricultural question answering, classification, and generation tasks. The dataset includes 1,152 steps across 34 agricultural items and is provided in JSON and CSV formats. It is suitable for NLP applications as well as other machine learning tasks such as temporal reasoning, procedural modeling, and data-to-text generation.