LexRAG
收藏资源简介:
LexRAG是由清华大学知识工程实验室提出的首个针对多轮法律咨询对话中检索增强生成(RAG)系统的基准数据集。该数据集包含1013个经过法律专家精心标注的多轮对话样本,每个样本包含五轮逐步提问,以及17228篇候选法律文章。数据集覆盖了包括民法、刑法、合同法、知识产权法等多个法律领域的条文。LexRAG旨在为法律领域中的RAG系统提供一个标准化的评估平台,以促进RAG在法律咨询系统中的高质量应用。
LexRAG is the first benchmark dataset for retrieval-augmented generation (RAG) systems in multi-turn legal consultation dialogues, proposed by the Knowledge Engineering Laboratory of Tsinghua University. This dataset contains 1013 multi-turn dialogue samples meticulously annotated by legal experts. Each sample includes five rounds of progressive questions, along with 17,228 candidate legal articles. The dataset covers legal provisions across multiple legal domains including civil law, criminal law, contract law, intellectual property law, etc. LexRAG aims to provide a standardized evaluation platform for RAG systems in the legal field, so as to promote high-quality applications of RAG in legal consultation systems.




