USClassActions

USClassActions数据集是由Darrow AI Ltd.精心策划，专注于美国集体诉讼案件的法律判决预测。该数据集包含10,800个案件，每个案件均标注了胜诉或败诉的二元结果。与多数其他法律判决预测数据集不同，USClassActions不仅来自采用普通法系的美国，而且专注于确保众多通常代表性不足的个人获得正义的重要类型投诉。数据集的创建过程中，研究人员手动审查了数百个来自不同法院和州的案件，以构建基于规则的正则表达式提取系统，从而从每个投诉中提取原告的事实和指控。USClassActions数据集的应用领域在于通过早期法律判决预测，使原告能够仅将有价值的案件带上法庭，同时被告更有可能更快地解决这些案件，从而支持联合国可持续发展目标16，即促进和平、包容的社会，为所有人提供司法救助，并建立有效、负责任和包容的机构。

The USClassActions dataset, carefully curated by Darrow AI Ltd., focuses on legal judgment prediction for U.S. class action lawsuits. It includes 10,800 cases, each annotated with a binary outcome indicating whether the plaintiff won or lost the case. Unlike most existing legal judgment prediction datasets, USClassActions not only originates from the U.S. common law system but also focuses on high-priority complaint types aimed at securing justice for a large number of typically underrepresented individuals. During the dataset's development, researchers manually reviewed hundreds of cases from various courts and states to build a rule-based regular expression extraction system, which extracts the plaintiff's factual allegations and claims from each complaint. The application of the USClassActions dataset enables plaintiffs to file only meritorious lawsuits through early legal judgment prediction, while allowing defendants to resolve such cases more efficiently, thereby supporting UN Sustainable Development Goal 16: to promote peaceful and inclusive societies, provide access to justice for all, and build effective, accountable and inclusive institutions.