Sentiment Analysis for Financial News Using RNN-LSTM Network

Abstract

Identifying financial sector sentiment, primarily through the financial news, is crucial in financial investment decisions. Financial Sentiment Analysis (FSA) significantly affects the secondary market and provides a significant contribution. Long Short-Term Memory (LSTM) is a deep learning model, especially Recurrent Neural Network (RNN), a reasonably popular model designed for long-term constraints. Various methods have been proposed in previous studies, but the performance generated by the model in previous studies is still below 90%, so it has the opportunity to be improved. This study aims to present a sentiment analysis model by implementing the RNN-LSTM. The results showed that the model used for sentiment analysis of financial news reached 92.23% precision, 91.54% accuracy, 90.99% recall, and an f1 score of 91.61%. The model we built is helpful for understanding trends and opinions for making financial and other investment decisions.