摘 要
随着信息技术的迅猛发展，大数据技术在金融领域的应用日益广泛，尤其在提升上市公司财务信息透明度方面展现出巨大潜力。本文旨在探讨大数据技术如何影响上市公司财务舞弊的识别效率与准确性，揭示其在实践中的应用价值与局限性。研究采用定量分析与案例研究相结合的方法，选取2010年至2022年间A股上市公司的公开数据，构建多元逻辑回归模型，并引入文本挖掘与异常交易监测等大数据手段进行实证检验。结果表明，融合多源异构数据的大数据分析框架能够显著提高财务舞弊识别的敏感度与特异性，尤其在识别隐蔽性强、形式复杂的舞弊行为方面具有明显优势。此外，研究还发现，基于机器学习算法构建的预警模型在预测潜在舞弊风险方面表现出良好的稳定性与前瞻性。本文的创新点在于系统整合了结构化与非结构化数据资源，提出了一个多维度、动态化的财务舞弊识别框架，并验证了大数据技术在该领域的有效性与可拓展性。研究结果为监管机构、审计机构及投资者提供了新的技术工具与决策支持路径，对提升资本市场治理水平具有重要的理论与现实意义。
关键词：大数据技术；财务舞弊识别；多源异构数据；机器学习预警模型；财务信息透明度
Abstract
With the rapid development of information technology, big data techniques have been increasingly applied in the financial sector, demonstrating significant potential particularly in enhancing the transparency of financial information for listed companies. This paper aims to explore how big data technologies influence the efficiency and accuracy of identifying financial fraud in listed companies, revealing their practical application value and limitations. The study employs a combination of quantitative analysis and case research methods, selecting publicly available data from A-share listed companies between 2010 and 2022 to construct a multiple logistic regression model, while incorporating empirical tests using big data techniques such as text mining and anomaly transaction monitoring. Results indicate that an integrated big data analytical fr amework combining multi-source heterogeneous data can significantly improve the sensitivity and specificity of financial fraud detection, especially in identifying fraudulent behaviors characterized by high concealment and complex forms. Furthermore, the study finds that early-warning models based on machine learning algorithms demonstrate strong stability and foresight in predicting potential fraud risks. The innovation of this paper lies in the systematic integration of structured and unstructured data resources, proposing a multidimensional and dynamic fr amework for detecting financial fraud and validating the effectiveness and scalability of big data techniques in this field. The findings provide regulatory authorities, auditing institutions, and investors with new technical tools and decision-support pathways, contributing both theoretically and practically to improving governance standards in capital markets.
Keywords: Big Data Technology; Financial Fraud Detection; Multi-source Heterogeneous Data; Machine Learning Early Warning Model; Financial Information Transparency
目 录
1绪论 1
1.1研究背景与现实意义 1
1.2国内外研究现状综述 1
1.3研究内容与方法设计 1
2大数据技术在财务舞弊识别中的应用基础 2
2.1上市公司财务舞弊的主要特征 2
2.2大数据技术的核心能力分析 2
2.3财务舞弊识别的传统方法局限性 3
2.4大数据技术对传统识别方式的改进 3
3大数据驱动下的财务舞弊识别模型构建 3
3.1数据来源与处理流程设计 3
3.2关键变量选取与指标体系建立 4
3.3机器学习算法在舞弊识别中的应用 4
3.4模型有效性评估与优化路径 5
4大数据技术提升舞弊识别效率的实证分析 5
4.1样本选择与数据描述 5
4.2实证模型设定与检验方法 5
4.3实证结果与分析讨论 6
4.4技术应用对企业治理的影响 6
5大数据环境下财务监管机制的优化路径 7
5.1监管机构的技术适应性挑战 7
5.2基于大数据的动态预警系统构建 7
5.3制度建设与技术协同发展的策略 8
5.4风险防控与信息披露机制完善 8
结论 8
参考文献 10
致 谢 11
