EDGAR-CORPUS

本研究使用的数据集为EDGAR-CORPUS，这是一个包含美国所有公开交易公司25年10-k报告的新经济数据集。数据集通过与UCLA-LoPucki破产研究数据库（BRD）匹配，利用中央索引键识别公司，获取破产预测所需的目标信息。BRD记录了自1997年以来美国破产法第7章和第11章的所有案件，并每月更新。研究限制了10-k报告至第7节“管理层讨论与分析”，该部分提供了公司对过去财务年度的业务结果的视角，并包含了可能显著影响公司的风险和不确定性。数据集的应用领域是预测破产，旨在通过文本数据预测企业未来的财务困境。

The dataset used in this study is EDGAR-CORPUS, a novel economic dataset containing 25-year worth of 10-K reports from all publicly traded companies in the United States. The dataset is matched with the UCLA-LoPucki Bankruptcy Research Database (BRD), where companies are identified using the Central Index Key (CIK) to obtain the target information required for bankruptcy prediction. The BRD documents all cases filed under Chapter 7 and Chapter 11 of the U.S. Bankruptcy Code since 1997 and is updated monthly. This study restricts the 10-K reports to their Item 7, Management's Discussion and Analysis (MD&A), which provides the company's perspective on the business outcomes of the past fiscal year and includes risks and uncertainties that may significantly impact the firm. The application scenario of this dataset is bankruptcy prediction, aiming to forecast the future financial distress of enterprises using textual data.