A quantum-like zero-shot approach for sentiment analysis in finance

Sentiment analysis has become an indispensable tool across various domains, including political communication, marketing analytics, and finance. In the financial sector, sentiments such as confidence or fear play a pivotal role in shaping market dynamics, influencing supply and demand, and precipitating significant price fluctuations. The timely extraction of sentiment from sources like financial news articles and social media posts is crucial for devising informed investment strategies. Despite its importance, current approaches to sentiment analysis in finance have not fully harnessed the potential of emerging technologies, such as large language models (LLMs) and quantum computing models. These cutting-edge technologies have demonstrated remarkable success in other areas of natural language processing (NLP), but their application in financial sentiment analysis remains limited. This paper aims to bridge this gap by introducing a novel quantum model for text representation, specifically designed for the financial domain. Our approach integrates this quantum representation with a vector-based representation generated by a pre-trained LLM specialized in finance. The resulting fusion of these two representations yields a more comprehensive input for an LLM, leading to a significant enhancement in the accuracy of the sentiment analysis task. We evaluate our approach through extensive experimental tests on a publicly available dataset and demonstrate that our quantum-based method outperforms state-of-the-art models on a widely recognized financial sentiment analysis benchmark. Our results highlight the potential of integrating quantum computing principles with traditional NLP methods for more accurate and effective sentiment analysis in finance.