The 21st Text Analytics Symposium, vol.124, no.173, NLC2024-15, pp. 81-86, Sep. 3, 2024
The 21st Text Analytics Symposium
This study aims to evaluate the sentiment of financial texts using large language models (LLMs) and to determine whether LLMs exhibit firm-specific biases in sentiment assessment. In particular, we investigate the impact of general knowledge about firms on the sentiment measurement of texts by LLMs. First, we compare the sentiment scores of performance-related texts provided by LLMs when the firm name is explicitly included in the prompt versus when it is not. Specifically, we define and quantify firm-specific bias as the difference in these scores. Next, we construct an economic model to theoretically evaluate the impact of sentiment bias on investor behavior. Finally, we conduct an empirical analysis using actual financial text data to examine the extent to which sentiment bias influences both corporate characteristics and stock performance. The findings of this study offer important insights into how sentiment bias affects price formation in financial markets.
large language model; sentiment analysis; bias; financial text mining;
@inproceedings{Nakagawa2024-texta,
title={{Evaluating company-specific bias in financial sentiment analysis using large language models [in Japanese]}},
author={Kei Nakagawa and Masanori Hirano and Yugo Fujimoto},
booktitle={The 21st Text Analytics Symposium},
volume={124},
number={173, NLC2024-15},
pages={81-86},
url={https://ken.ieice.org/ken/paper/20240903bc4F/},
year={2024}
}