Sentiment classification from financial content posts on social media

In today's digital era, social media has become an important data source for understanding investor behavior and market trends in financial markets. In this study, Twitter data and stock prices were used to examine the impact of social media content on financial markets, and the potential use of sentiment-based analysis in influencing the market was evaluated. A total of 10,505 tweets tagged with #xu100 and #bist100 were collected from July 2023 to July 2024 and analyzed along with features such as comment count, retweets, likes, views, time, date, and stock price. Using Natural Language Processing (NLP) techniques, each tweet was classified as positive, negative, or neutral. The results indicated that tweets with positive sentiment were associated with increases in stock prices, while tweets with negative sentiment showed a tendency toward declines. Three-class classification was conducted using LSTM and GRU models; although training accuracy was high in three-class classification models, an overfitting problem was observed in test accuracy. Model performance was evaluated using precision, recall, and F1 score metrics, and difficulties were identified in recognizing negative classes. These findings highlight the potential of sentiment classification in tracking financial market movements and provide suggestions for future model improvements.