基于人工智能技术的管理层语调识别及其有效性——来自对企业长期投资预测效果的证据

张广婷1; 顾淳晖2; 洪康隆3

基于人工智能技术的管理层语调识别及其有效性——来自对企业长期投资预测效果的证据

引用本文：张广婷1，顾淳晖2，洪康隆3.基于人工智能技术的管理层语调识别及其有效性——来自对企业长期投资预测效果的证据[J].财经理论与实践,2025,(4):43-50

摘要点击次数: 248

全文下载次数: 315

作者	单位
张广婷1，顾淳晖2，洪康隆3	（1. 上海社会科学院世界经济研究所，上海 200020；2. 复旦大学经济学院，上海 200433；3. 中证指数有限公司，上海 200135）

中文摘要:以上市公司管理层讨论与分析文本为研究样本，研究谷歌公司的BERT人工智能模型能否识别上市公司高管的真情实感。研究发现, BERT模型度量下的上市公司管理层语调能反映管理层对于企业未来发展前景的态度;管理层预期任期越长，持股比例越高，内部控制水平越高时，其语调积极程度越高对企业长期投资的预测效果越好；进一步的对比研究发现，相比以往的词袋法，BERT模型的识别效果更好。鉴于此，投资者利用人工智能工具能更有效地对上市公司管理层语调进行情感分析，并预测公司的未来投资行为和财务绩效。

中文关键词:人工智能管理层语调 BERT模型长期投资

Management Intonation Recognition and Effectiveness Based on Artificial Intelligence Technology— Evidence from the Effect of Forecasting Long-term Investment by Firms

Abstract:Based on the management discussion and analysis of listed companies, this paper studies whether Google’s BERT artificial intelligence model can identify the true feelings of executives in listed companies. The findings are as follows: the management intonation measured by BERT AI model can reflect the management’s attitude towards the future development prospects of the enterprise. The longer the expected tenure of the management, the higher the shareholding ratio and the higher the level of internal control, the stronger the positive intonation of the management is in predicting the long-term investment. Further research finds that compared with the previous bag of words method, BERT model has a better recognition effect on management intonation. Therefore, investors using AI tools are more effective at sentiment analysis of management intonation and predicting future investment behavior and financial performance of the company.

keywords:artificial intelligence management intonation BERT model long-term investment

查看全文 查看/发表评论 下载pdf阅读器