KBL
收藏资源简介:
KBL数据集是由首尔大学、LBox和汉阳大学的研究人员开发的,专门用于评估大型语言模型对韩语法律语言的理解能力。该数据集包含7个法律知识任务(510个示例)、4个法律推理任务(288个示例)以及韩国律师考试的多个选择题(2510个示例),总计3308个示例。数据集的内容来源于韩国的判例、法令、律师考试等多个权威来源,确保了数据的高质量和实用性。数据集的创建过程与法律专业人士紧密合作,确保了任务的设计和质量保证。KBL数据集主要应用于法律领域的自然语言处理任务,旨在提升大型语言模型在处理韩语法律文本时的准确性和实用性。
The KBL dataset was developed by researchers from Seoul National University, LBox, and Hanyang University, specifically constructed to evaluate the Korean legal language comprehension capabilities of large language models (LLMs). This dataset encompasses 7 legal knowledge tasks (510 examples), 4 legal reasoning tasks (288 examples), and a collection of multiple-choice questions from the Korean Bar Exam (2510 examples), totaling 3308 examples. The content of the dataset is derived from multiple authoritative South Korean sources, including judicial precedents, statutory laws, and bar examinations, ensuring high data quality and practical utility. The development of the dataset was carried out in close collaboration with legal professionals to guarantee both task design and quality assurance. The KBL dataset is primarily applied to natural language processing tasks in the legal domain, aiming to enhance the accuracy and practical applicability of large language models when processing Korean legal texts.

- 1Developing a Pragmatic Benchmark for Assessing Korean Legal Language Understanding in Large Language Models首尔大学 LBox 汉阳大学 · 2024年



