arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~
arXiv 2404.05281cs.CL
收藏

用于财务年报特征提取的多任务学习

Multi-Task Learning for Features Extraction in Financial Annual Reports

Jožef Stefan Institute(约热·斯特凡研究所);School of Economics and Business, University of Ljubljana(卢布尔雅那大学经济与商学院)

Syrielle Montariol, Matej Martinc, Andraž Pelicon, Senja Pollak, Boshko Koloski, Igor Lončarski, Aljoša Valentinčič

AI总结:

本文提出多任务金融文本分类方法,通过显式融合辅助任务预测提升情感、客观性、前瞻句及 ESG 检测效果,并用于分析 FTSE350 年报文本特征与 ESG 评分的关联。

摘要:

在评估公司的各类绩效指标时,关注重点正从严格的财务(定量)公开披露信息转向定性(文本)信息。这类文本数据能够提供有价值的弱信号,例如通过文体特征,补充有关财务绩效或环境、社会与治理(ESG)标准的定量数据。在本研究中,我们采用多种多任务学习方法进行金融文本分类,重点关注金融情感、客观性、前瞻性句子预测以及 ESG 内容检测。我们提出了不同方法,以结合在不同任务上联合训练所提取的信息;其中表现最佳的方法凸显了在多任务训练过程中,将辅助任务预测显式作为特征加入最终目标任务所产生的积极效果。随后,我们使用这些分类器从 FTSE350 公司的年报中提取文本特征,并研究 ESG 定量评分与这些特征之间的联系。

英文摘要:

For assessing various performance indicators of companies, the focus is shifting from strictly financial (quantitative) publicly disclosed information to qualitative (textual) information. This textual data can provide valuable weak signals, for example through stylistic features, which can complement the quantitative data on financial performance or on Environmental, Social and Governance (ESG) criteria. In this work, we use various multi-task learning methods for financial text classification with the focus on financial sentiment, objectivity, forward-looking sentence prediction and ESG-content detection. We propose different methods to combine the information extracted from training jointly on different tasks; our best-performing method highlights the positive effect of explicitly adding auxiliary task predictions as features for the final target task during the multi-task training. Next, we use these classifiers to extract textual features from annual reports of FTSE350 companies and investigate the link between ESG quantitative scores and these features.

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。

↑