LEVEN
收藏资源简介:
LEVEN是一个大规模的中文法律事件检测数据集,由清华大学创建,包含8,116份法律文档和150,977个人工标注的事件提及,涵盖108种事件类型。该数据集不仅关注与指控相关的事件,还包括对案件理解至关重要的通用事件,这些在现有数据集中常被忽视。LEVEN是目前最大的法律事件检测数据集,其数据规模是其他数据集的数十倍,旨在显著推动法律事件检测方法的训练和评估。此外,LEVEN还展示了法律事件检测对于下游法律人工智能应用的基础性作用,如法律判决预测和类似案件检索,表明法律事件检测能够提供细粒度信息,从而作为法律人工智能的基本处理过程。
LEVEN is a large-scale Chinese legal event detection dataset developed by Tsinghua University. It contains 8,116 legal documents and 150,977 manually annotated event mentions, covering 108 distinct event types. This dataset not only focuses on charge-related events, but also includes general events that are critical for case understanding but often overlooked in existing datasets. LEVEN is currently the largest legal event detection dataset, with a scale dozens of times that of other datasets, aiming to significantly advance the training and evaluation of legal event detection methods. Furthermore, LEVEN demonstrates the fundamental role of legal event detection in downstream legal artificial intelligence applications such as legal judgment prediction and similar case retrieval, showing that legal event detection can provide fine-grained information and thus serve as a fundamental processing step for legal AI.

- 1LEVEN: A Large-Scale Chinese Legal Event Detection Dataset清华大学 · 2022年



