Contract Discovery

Contract Discovery数据集由波兰华沙的Applica.ai机构创建，专注于从法律文档中提取条款，以支持自动化合同发现任务。该数据集包含约2500个条款，来源于美国EDGAR数据库的债券发行说明书和保密协议文档，以及英国慈善机构年度报告。数据集旨在通过提供自然语言流，缺乏正式结构，模拟真实世界的使用场景，支持查询示例场景中的跨文档条款识别。该数据集适用于少样本学习设置，旨在通过多示例查询系统提高法律信息检索的效率和准确性。

The Contract Discovery dataset was created by Applica.ai, an institution based in Warsaw, Poland, which focuses on extracting clauses from legal documents to support automated contract discovery tasks. It contains approximately 2,500 clauses sourced from bond offering prospectuses and confidentiality agreement documents in the U.S. EDGAR database, as well as annual reports of UK charities. The dataset is designed to simulate real-world usage scenarios by providing unstructured natural language streams, supporting cross-document clause identification in query example scenarios. This dataset is suitable for few-shot learning settings, aiming to improve the efficiency and accuracy of legal information retrieval through multi-example query systems.