Mobilizing Text As Data
综述了文本分析方法在金融和会计领域的应用，涵盖公司披露、财报电话会、投资者关系和社交媒体等场景，讨论了关键词列表、词嵌入和大语言模型等方法的优缺点，并指导如何选择方法、验证指标和报告结果。
Textual analysis methods have become increasingly popular and powerful tools for researchers in finance and accounting to extract meaningful information from unstructured text data. This paper surveys the recent applications of these methods in various domains, such as corporate disclosures, earnings calls, investor relations, and social media. It also discusses the advantages and challenges of different textual analysis methods, such as keyword lists, pattern-based sequence classification, word embedding, and other large language models. We provide guidance on how to choose appropriate methods, validate text-based measures, and report text-based evidence effectively. We conclude by suggesting some promising directions for future research using text as data.