Textual Analysis in Economics and Finance

经济与金融文本分析

• 批准号: 286229948
• 负责人: Dr. Nikolas Breitkopf, since 10/2016
• 金额: --
• 依托单位: Institut für Finance und Banking
• 依托单位国家: 德国
• 项目类别: Scientific Networks
• 财政年份: 2016
• 资助国家: 德国
• 起止时间: 2015-12-31 至 2016-12-31
• 项目状态: 已结题
• 来源: https://gepris.dfg.de/gepris/projekt/286229948?language=en
• 关键词: Textual; Analysis; Economics; Finance

Practitioners and researchers face the challenge of having to extract information from the vast amount of unstructureddata available today. This data typically takes the form of text written in natural language, such ascorporate filings, transcripts of earnings calls, news articles, twitter feeds, etc.. Due to its size, it is infeasibleto manually process this data. Instead, computer-based methods of textual analysis are required. Textualanalysis, i.e. automated extraction of information from text, requires fundamentally different approaches comparedto analyzing structured data, such as tabulated time-series data. Academics in finance have startedto adopt methods of textual analysis from linguistics and machine learning for their research. For instance,textual analysis is being applied to extract the tone or complexity of financial texts and analyze their effect onfirm performance and investor behavior.Despite these first successful attempts, textual analysis is not yet part of the standard toolset in financialeconomics research. This is due to methodological issues of this relatively new discipline, and due to the factthat textual analysis has yet to be proven useful in the various fields of financial economics. The scientificnetwork aims to address these challenges. To this end, three areas of a lack of existing research have beenidentified:1. Testing the ability of existing methods to select and classify information in financial texts.2. Application of textual analysis in the context of capital markets and corporate finance applications.3. Application of textual analysis in the international context, i.e. application to text written in foreign languages.

从业者和研究人员面临的挑战是，必须从今天可用的海量非结构化数据中提取信息。这些数据通常采用自然语言编写的文本形式，如公司备案文件、收益电话会议记录、新闻文章、Twitter提要等。由于其大小，手动处理这些数据是不可行的。取而代之的是基于计算机的文本分析方法。文本分析，即从文本中自动提取信息，需要与分析结构化数据(如表格时间序列数据)完全不同的方法。金融领域的学者已经开始采用语言学和机器学习中的文本分析方法进行研究。例如，文本分析正被应用于提取金融文本的基调或复杂性，并分析它们对公司业绩和投资者行为的影响。尽管有这些第一次成功的尝试，但文本分析还不是金融经济学研究的标准工具集的一部分。这是由于这一相对较新的学科的方法论问题，以及文本分析尚未被证明在金融经济学的各个领域有用的事实。科学网络旨在应对这些挑战。为此，发现了现有研究不足的三个方面：1.检验现有方法对金融文本中的信息进行选择和分类的能力。文本分析在资本市场和公司融资应用中的应用。文本分析在国际背景下的应用，即对用外语编写的文本的应用。