无线电通信技术2025，Vol.51Issue(3)：501-510,10.DOI:10.3969/j.issn.1003-3114.2025.03.009
基于多模态财务大数据的智能分析预测模型
Intelligent Analysis and Prediction Model Based on Multi Modal Financial Big Data
摘要
Abstract
Recently,there has been growing interest in text sentiment analysis and speech emotion recognition in the financial do-main,as they have the potential to capture the intentions and opinions of corporate stakeholders,such as managers and investors.By incorporating text sentiment,companies have achieved significant performance improvements in predicting financial performance.How-ever,considering only text sentiment is far from sufficient to understand the emotional states of managers and to predict financial crises.An intelligent analysis and prediction model based on multimodal financial big data called Multi Modal Financial Intelligent Model(MMFIM)is proposed,which integrates multimodal data including voice sentiment features,text sentiment features,and traditional fi-nancial features to address the aforementioned issues.The MMFIM first identifies the emotional states of managers through a Convolu-tional Neural Network(CNN)model and extracts text sentiment features by using the FinBERT model.Then,it fuses sentiment infor-mation with traditional financial measurements to achieve more accurate predictions of financial crises.The proposed model is validated on data from 1 278 earnings call conferences of the 40 largest companies(mark-to-market)in the United States.Experimental results show that the proposed MMFIM has higher prediction accuracy compared to existing prediction models.关键词
多模态/语音情感识别/文本情感分析/卷积神经网络/FinBERT/财务危机预测Key words
multi modal/speech emotion recognition/text sentiment analysis/CNN/FinBERT/financial distress prediction分类
信息技术与安全科学引用本文复制引用
王伯平,王邦平..基于多模态财务大数据的智能分析预测模型[J].无线电通信技术,2025,51(3):501-510,10.基金项目
四川省科技计划资助(2024YFFK0413) Sichuan Science and Technology Program(2024YFFK0413) （2024YFFK0413）
