从数字到文字:使用 ECL 数据集的多模态破产预测
From Numbers to Words: Multi-Modal Bankruptcy Prediction Using the ECL Dataset
AI总结:
本文构建包含企业10-K文本与数值信息的多模态破产数据集 ECL,评估经典与神经模型,揭示模态互补性及二元标签局限,并考察 GPT 类 LLM 的摘要与零样本预测能力。
摘要:
本文提出 ECL,这是一个新颖的多模态数据集,包含来自企业 10-K 文件的文本和数值数据,以及相关的二元破产标签。此外,我们利用该数据集开发并严格评估了若干经典和神经网络破产预测模型。研究结果表明,每种数据模态所含信息在破产预测中具有互补性。我们还发现,二元破产预测目标无法使模型区分下一年破产与将在以后年份导致破产的不健康财务状况。最后,我们探索了 LLM 在该任务背景下的应用。结果显示,基于 GPT 的模型可用于从文本数据中提取有意义的摘要,但零样本破产预测结果较差。访问和更新该数据集或复现我们实验所需的全部资源均可在 github.com/henriarnoUG/ECL 获取。
英文摘要:
In this paper, we present ECL, a novel multi-modal dataset containing the textual and numerical data from corporate 10K filings and associated binary bankruptcy labels. Furthermore, we develop and critically evaluate several classical and neural bankruptcy prediction models using this dataset. Our findings suggest that the information contained in each data modality is complementary for bankruptcy prediction. We also see that the binary bankruptcy prediction target does not enable our models to distinguish next year bankruptcy from an unhealthy financial situation resulting in bankruptcy in later years. Finally, we explore the use of LLMs in the context of our task. We show how GPT-based models can be used to extract meaningful summaries from the textual data but zero-shot bankruptcy prediction results are poor. All resources required to access and update the dataset or replicate our experiments are available on github.com/henriarnoUG/ECL.
机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。