[4Xin2-11] Inference-based sentiment analysis of financial text considering business profile with large language model
Keywords:Large Language Model, Financial Text Mining, Sentiment Analysis, Prompt Engineering, Natural Language Processing
Sentiment analysis plays a critical role in financial text mining. However, conventional approaches typically focus on texts that directly describe positive or negative impacts on financial performance. Considering practical investment decisions, it's necessary to extract sentiment from texts that may not explicitly express sentiment, taking into account the background context. Particularly, such an inferring sentiment becomes even more important for small and mid-cap companies, which rarely receive news coverage.
Therefore, in this study, we utilize the inference capabilities of a Large Language Model (LLM) to tackle an inference-based sentiment analysis task. Specifically, we input the business profile of a company as background context and infer the impact of a particular major event on the company performance, subsequently assigning sentiment. We conduct the experiments to assess the practical usefulness of the model's output.
Therefore, in this study, we utilize the inference capabilities of a Large Language Model (LLM) to tackle an inference-based sentiment analysis task. Specifically, we input the business profile of a company as background context and infer the impact of a particular major event on the company performance, subsequently assigning sentiment. We conduct the experiments to assess the practical usefulness of the model's output.
Please log in with your participant account.
» Participant Log In