Real-Time Analysis of Stock Market Prediction Through Social Networks Data

Researchers are expending significant effort in devising precise methodologies and strategies to predict movements in the stock market. The acquisition of information has now become a critical component in evaluating the actions and behaviors of individuals. Utilizing Deep Learning (DL) and analysis of sentiment on data gathered from microblogging assistance has emerged as a commonly employed approach for predicting stock market movements. Within this scholarly article, we propose a methodology that employs sentiment analysis on Twitter and stock-related tweets to predict stock movements. Applying sentiment analysis to tweets, we implemented two DL models, namely CNN and LSTM. The GloVe algorithm computes word embeddings by utilizing a co-occurrence matrix among words. The most favorable outcomes were attained when employing tweets sourced from Twitter in conjunction with CNN and LSTM. The highest level of accuracy achieved was 88.34%, while the top precision reached 85.26%. Furthermore, CNN demonstrated the utmost accuracy, specifically 83.56%, when utilizing Stock Tweets with a balanced data set.