Large Language Models in Accounting and Financial Research: A Review of Applications in Accounting-related Text Analysis

With the rapid advancement of large language models (LLM) technologies, new technical support and research paradigms have emerged for traditional accounting and finance studies. Despite these opportunities, scholars in the fields of accounting and finance often encounter challenges when navigating the extensive and complex domain knowledge of large language models, as well as the ever-evolving academic literature. To address this gap, this paper conducts a qualitative survey on the applications of large language models in early accounting and finance researches. This paper is structured into three main sections. First, it delves into the definition, underlying principles, and developmental trajectory of large language models. Second, it synthesizes the latest research on large language models applications in accounting and finance, categorizing these studies into three major domains, including sentiment analysis, report analysis and practical works. Finally, the paper highlights emerging trends and potential research directions, aiming to provide a comprehensive guide for future scholarly exploration in this dynamic field. Although it still has many shortcomings such as high cost and black box of process, the application of large language modeling provides a new technical support for text analysis in accounting and financial research, and also improves the actual efficiency of accounting and auditing.