基于BERTopic模型的企业ESG政策文本主题识别与趋势分析

摘要：

【目的】紧跟时事政策，探究发展趋势，更好地为企业提供参考价值，也为社会公众更整体地了解政策变化。【方法】采用BERTopic模型、数据搜集、文本分析等方法，对近几年各个单位出台的关于企业ESG相关政策进行总结分类和评价，审视主要方向并提出不足和建议。【结果】上市公司环境战略、内部环境管理制度、企业信息披露评估框架、绿色债券信息披露、统一国内绿色债券项目认定标准、ESG信息披露的基本架构、定期披露公司信息的义务方面较好，但研究对象存在追责机制不健全、缺少ESG评级质量标准、ESG审计不足的问题。【结论】提出了需要完善追责机制、统一评级质量标准和提升ESG审计水平的建议，最终有效遏制“漂绿”乱象的发生，提升ESG信息质量与公信力，实现长期发展。

Abstract:

[Objective] The study aims to keep up with current policies, explore development trends, provide better reference for enterprises, and help the public have a more comprehensive understanding of policy changes. [Methods] This paper uses the BERTopic model, data collection, text analysis and other methods to summarize, classify and evaluate the ESG related policies issued by various units in recent years, examine the main directions, reveal shortcomings and give suggestions. [Results] The results indicate that listed companies have good environmental strategies, internal environmental management systems, corporate information disclosure evaluation frameworks, green bond information disclosure, unified domestic green bond project recognition standards, basic framework for ESG information disclosure, and obligation to regularly disclose company information. However, the research subjects have problems such as incomplete accountability mechanisms, lack of ESG rating quality standards, and insufficient ESG audits. [Conclusion] This paper proposes suggestions for improving accountability mechanisms, unifying rating quality standards, and enhancing ESG audit capabilities, so as to effectively curb the occurrence of “greenwashing” chaos, enhance ESG information quality and credibility, and achieve long-term development.

引用格式：孟志华，耿嘉璘.基于BERTopic模型的企业ESG政策文本主题识别与趋势分析[J]. 标准科学，2026（3）：36-44.

基金项目：本文受甘肃省哲学社会科学规划项目“化债背景下的特许经营项目协同治理审计机制优化研究”（项目编号：2024YB083）资助。

作者简介：孟志华，硕士，副教授，硕士研究生导师，研究方向为审计理论与实务。耿嘉璘，硕士研究生，研究方向为社会审计。