计算机科学

计算机科学 ›› 2026, Vol. 53 ›› Issue (6A): 250500073-14.doi: 10.11896/jsjkx.250500073

• 人工智能 • 上一篇    下一篇

基于市场反馈监督的大语言模型金融文本情绪分析方法研究

张永宇1,2, 郭晨娟1, 费雪琴3, 黎峰4   

  1. 1 华东师范大学数据科学与工程学院 上海 200062
    2 恒生电子股份有限公司 杭州 310052
    3 浙江宏信安检测认证技术有限公司 杭州 311400
    4 证通股份有限公司 上海 200131
  • 出版日期:2026-06-16 发布日期:2026-06-12
  • 通讯作者: 郭晨娟(cjguo@dase.ecnu.edu.cn)
  • 作者简介:(yongyu.zhang@gmail.com)

Study on Financial Text Sentiment Analysis Method Based on Large Language Models with Market Feedback Supervision

ZHANG Yongyu1,2, GUO Chenjuan1, FEI Xueqin3, LI Feng4   

  1. 1 School of Data Science & Engineering,East China Normal University, Shanghai 200062,China
    2 Hundsun Technologies Inc.,Hangzhou 310052,China
    3 Zhejiang X Testing and Certification Technology Co., Ltd., Hangzhou 311400,China
    4 E-Capital Transfer Co., Ltd.,Shanghai 200131,China
  • Published:2026-06-16 Online:2026-06-12
  • About author:ZHANG Yongyu,born in 1979,master,senior engineer,is a member of CCF(No.R9350M).Hismain research interests include financial time series prediction and multimodal tasks.
    GUO Chenjuan,born in 1982,Ph.D,professor,Ph.D supervisor.Her main research interests include data management and data analysis.

PDF (PC)

43

摘要： 在金融市场,市场情绪对于资产价格以及波动所产生的影响是极为深远的,大语言模型尽管给予了金融文本情绪分析一定的机遇,然而当前的研究却存在着诸多问题,比如难以有效处理金融文本所有的专业性、动态性,并且与市场反应的一致性较差。对此,构建了一种创新的金融文本情绪分析体系,将多种大语言模型的优势融合,同时结合知识图谱提高技术以及思维链技术来对混合语言模型框架进行优化,运用多时间窗口与动态权重相结合的滑动分析方法,构建起市场指标评估体系并且开发出自适应动态更新算法,以此强化市场反馈监督机制。通过实证分析可发现,该体系在情绪分析的准确性方面表现十分出色,与市场反应的一致性较高,明显要优于对比模型。本研究为金融市场研究以及投资决策提供了全新的视角和工具,在金融领域有重要的理论意义与实践意义。

关键词: 金融文本情感分析, 大语言模型, 市场反馈监督, 多模型融合, 知识图谱

Abstract: In the financial market,market sentiment has a profound impact on asset prices and volatility.Although large language models bring opportunities for financial text sentiment analysis,current research still has issues such as difficulties in handling the professionalism and dynamics of financial texts and poor consistency with market reactions.This study constructs an innovative financial text sentiment analysis system.It integrates the advantages of multiple large language models,combines knowledge graph enhancement technology and chain-of-thought technology to optimize the hybrid language model framework.Moreover,it adopts a sliding analysis method that combines multiple time windows and dynamic weights,constructs a market index evaluation system,and develops an adaptive dynamic update algorithm to strengthen the market feedback supervision mechanism.Empirical analysis shows that this system performs excellently in the accuracy of sentiment analysis and has a high consistency with market reactions,significantly outperforming comparative models.This research provides new perspectives and tools for financial market research and investment decision-making,and holds great theoretical and practical significance in the financial field.

Key words: Financial text sentiment analysis, Large language models, Market feedback supervision, Multi-model fusion, Know-ledge graph

中图分类号: 

  • F224-39
[1] GHOSAL S,LIU C,ARUL APPAN A,et al.Large Language Models and Sentiment Analysis in Financial Markets:A Review,Datasets,and Case Study [J].IEEE Access,2024,12:134041-134061.
[2] JIANG F W,LIU Y M,MENG L C.Large language models, text sentiment and financial markets[J].Journal of Management World,2024,40(8):42-64.
[3] HUANG A H,WANG H,YANG A.FinBERT:A Large Language Model for Extracting Information from Financial Text.Contemporary Accounting Research,2022,39(1):559-595.
[4] WU S,OZAN R S,LU S,et al.BloombergGPT:A Large Language Model for Finance[J].arXiv:2304.01373,2023.
[5] LI X,CHEN C,WANG Y.Research on the relationship between financial text sentiment and market response based on multi-factor model[J].Financial Theory and Practice,2023(7):44-52.
[6] ZHU J,NIU Z L.Public opinion mining and analysis based on large language models[J].Financial Technology Times,2024(5):59-63.
[7] Techbeatly.Financial Sentiment Analysis:Large Language Model,RAG[EB/OL].https://www.techbeatly.com/financial-sentiment-analysis-large-language-model-rag/.
[8] SMITH J,JOHNSON A.Analyzing the Lagged Relationship between Financial Text Sentiment and Market Reactions Using Time-Series and LLM-based Approaches [J].Journal of Financial Research,2023,45(3):45-60.
[9] BROWN D,GREEN M.Dynamic Adjustment of LLM-based Financial Sentiment-Market Reaction Models using Reinforcement Learning [J].Computational Economics Review,2024,22(2):25-40.
[10] JIN Y,LI C Z.Evaluation and enlightenment of financial capabilities of AI large models:An empirical test of ChatGPT and domestic large models based on CPA examination[J].Finance and Accounting Monthly,2024,45(18):44-51.
[11] SHI Z H,LU M F.Application,challenges and countermeasures of generative artificial intelligence in financial service innovation[J].Financial Technology Times,2024,32(8):8-18.
[12] LUO S J.Financial large models:Application,risks and institutional responses[J].Financial Development Research,2024(6):70-78.
[13] ZHONG H,CHEN S S,WANG M.Research on the reasoning ability of large language models in the accounting domain[J].Finance and Accounting Monthly,2024,45(11):17-25.
[14] LIU Z X.Generative artificial intelligence empowering inclusive finance:Realistic foundation,key risk challenges and coping strategies[J].Frontiers,2024(6):86-93.
[15] LU M F.Research on the application principle,challenges and implementation path of large language models in the financial sector[J].Journal of Chongqing Technology and Business University(Social Sciences Edition),2024,41(4):1-12.
[16] DING X W,ZHAO X,LIU Z H,et al.ChatGPT in accurate public opinion judgment:Theoretical and application exploration driven by the paradigm of “big data,large models,big computing”[J].Media Observer,2024(3):45-61.
[17] GONG X Y,JIANG H.The impact of generative large models on commercial banks and relevant suggestions[J].New Finance,2023(10):32-37.
[18] ZHANG X F,ZHANG L P,YAN S,et al.Personalized learning recommendation based on knowledge graph and large language model collaboration[J].Journal of Computer Applications,2025,43(3):773-784.
[19] ZHANG C H,LUO J,TU X H,et al.Automatic generationmethod of corpus query based on large language model[J].Computer Engineering,2025,52(2):404-412.
[20] TAO J Y,XI X F,SHENG S L,et al.A review on enhancing reasoning ability of large language models with structured thinking prompting[J].Computer Engineering and Applications,2025,61(6):64-83.
[21] SHENG K,WANG Z Q.Synaesthetic metaphor analysis based on large language model and data augmentation[J].Journal of Computer Applications,2025,45(3):794-800.
[22] YANG Y X,WU J H,WU Y K,et al.Research on false health information identification method fusing knowledge graph and large language model[J].Information Studies:Theory & Application,2025,48(3):127-133.
[23] WANG H T,SHI Y K.Research on query expansion method based on large language model[J].Computer Technology and Development,2025,35(3):148-155.
[24] XU B.Application and prospect of generative artificial intelli-gence technology in commercial banks[J].Technology and Finance,2023(9):61-67.
[25] HOU P,LI Z X,ZHANG F,et al.Intelligent technologies and practices for financial data security governance[J].Chinese Journal of Network and Information Security,2023,9(3):174-187.
[26] ZHANG S Y,KUANG F J.Financial panic public opinion analysis and prediction based on LDA-BiLSTM[J].Software Guide,2023,22(10):79-83.
[27] LI B,GAO L X.The intelligent transformation of public opinion governance under the background of big data:Current situation,risks and countermeasures[J].China Editor Journal,2023 (5):4-10.
[28] CHEN F J,XIAO H Y.Research on network public opinion heat prediction based on improved ABC-BP model[J].Research on Library Science,2018(9):84-89,101.
[29] LIN P,HUANG W D.Research on the topic evolution path of network sudden events based on LDA model[J].Information Science,2014,32(10):20-23.
[30] ZHANG Z A,LIAO X.Key issues of public opinion guidance in a complex public opinion environment[J].Journalism Lover,2024(12):4-9.
[31] GUO X W.Research on semantic analysis algorithm of network public opinion based on long short-term memory network[J].Digital Technology and Application,2024,42(5):49-51.
Viewed
Full text


Abstract

Cited
  Shared   
  Discussed   
No Suggested Reading articles found!
渝ICP备16013121号-4
地址：重庆市两江新区洪湖西路18号    邮编：401121  
电话：023-63500828（编辑部） 023-67039612（会议/专题） 023-67039623（财务/发票）
E-mail：jsjkx12@163.com
本系统由北京玛格泰克科技发展有限公司设计开发