Large Language Models and Sentiment Analysis in Financial Markets: A Review, Datasets, and Case Study

被引：2

作者：

Liu, Chenghao ^{[1
]}

Arulappan, Arunkumar ^{[2
]}

Naha, Ranesh ^{[3
]}

Mahanti, Aniket ^{[1
,4
]}

Kamruzzaman, Joarder ^{[5
]}

Ra, In-Ho ^{[6
]}

机构：

[1] Univ Auckland, Sch Comp Sci, Auckland 1010, New Zealand

[2] VIT Univ, Sch Comp Sci Engn & Informat Syst, Vellore 632014, India

[3] Queensland Univ Technol, Sch Informat Syst, Brisbane, Qld 4000, Australia

[4] Univ New Brunswick, Dept Comp Sci, St John, NB E2K 5E2, Canada

[5] Federat Univ Australia, Ctr Smart Analyt, Melbourne, Vic 3806, Australia

[6] Kunsan Natl Univ, Sch Software, Gunsan 54150, South Korea

来源：

IEEE ACCESS | 2024年 / 12卷

关键词：

Sentiment analysis; Bitcoin; Analytical models; Predictive models; Large language models; Correlation; Quality assessment; Large language model; Bitcoin price; sentiment analysis; machine learning; market dynamics; INVESTOR SENTIMENT; INFORMATION; FRAMEWORK; CONTAGION; MEDIA;

D O I：

10.1109/ACCESS.2024.3445413

中图分类号：

TP [自动化技术、计算机技术];

学科分类号：

0812 ;

摘要：

This paper comprehensively examines Large Language Models (LLMs) in sentiment analysis, specifically focusing on financial markets and exploring the correlation between news sentiment and Bitcoin prices. We systematically categorize various LLMs used in financial sentiment analysis, highlighting their unique applications and features. We also investigate the methodologies for effective data collection and categorization, underscoring the need for diverse and comprehensive datasets. Our research features a case study investigating the correlation between news sentiment and Bitcoin prices, utilizing advanced sentiment analysis and financial analysis methods to demonstrate the practical application of LLMs. The findings reveal a modest but discernible correlation between news sentiment and Bitcoin price fluctuations, with historical news patterns showing a more substantial impact on Bitcoin's longer-term price than immediate news events. This highlights LLMs' potential in market trend prediction and informed investment decision-making.

引用

页码：134041 / 134061

页数：21

共 106 条

[51] ON INFORMATION AND SUFFICIENCY [J].

KULLBACK, S ;

LEIBLER, RA .

ANNALS OF MATHEMATICAL STATISTICS, 1951, 22 (01) :79-86

[52]

Lan Zhenzhong, 2020, INT C LEARNING REPRE

[53]

Laskar MTR, 2023, FINDINGS OF THE ASSOCIATION FOR COMPUTATIONAL LINGUISTICS, ACL 2023, P431

[54]

Le Scao Teven, 2022, arXiv

[55] Interpretable deep learning: interpretation, interpretability, trustworthiness, and beyond [J].

Li, Xuhong ;

Xiong, Haoyi ;

Li, Xingjian ;

Wu, Xuanyu ;

Zhang, Xiao ;

Liu, Ji ;

Bian, Jiang ;

Dou, Dejing .

KNOWLEDGE AND INFORMATION SYSTEMS, 2022, 64 (12) :3197-3234

[56]

Li Y, 2023, Arxiv, DOI arXiv:2309.03736

[57]

Liu YH, 2019, Arxiv, DOI [arXiv:1907.11692, 10.48550/arXiv.1907.11692, DOI 10.48550/ARXIV.1907.11692]

[58]

Liu Z, 2020, PROCEEDINGS OF THE TWENTY-NINTH INTERNATIONAL JOINT CONFERENCE ON ARTIFICIAL INTELLIGENCE, P4513

[59]

Malo P, 2013, Arxiv, DOI arXiv:1307.5336

[60] Good Debt or Bad Debt: Detecting Semantic Orientations in Economic Texts [J].

Malo, Pekka ;

Sinha, Ankur ;

Korhonen, Pekka ;

Wallenius, Jyrki ;

Takala, Pyry .

JOURNAL OF THE ASSOCIATION FOR INFORMATION SCIENCE AND TECHNOLOGY, 2014, 65 (04) :782-796

← 1 2 3 4 5 6 7 8 9 10 →