Published Paper Details:

Stock market forecasting is a challenging task due to its volatile nature and dependence on multiple factors, including historical trends, economic indicators, and market sentiment. Traditional models, such as statistical and machine learning approaches, primarily focus on historical price data, often neglecting the crucial role of public sentiment in influencing stock prices. This research introduces a hybrid model that integrates Long Short-Term Memory (LSTM) neural networks with Natural Language Processing (NLP)-based sentiment analysis to enhance the accuracy of stock market predictions. The proposed framework collects historical stock prices from financial platforms and textual data from news articles, financial reports, and social media. Sentiment analysis is performed using NLP techniques such as VADER, Text Blob, and BERT to classify market sentiment as positive, neutral, or negative. These sentiment scores are then incorporated into an LSTM-based deep learning model, which processes both numerical stock data and qualitative sentiment trends to generate more reliable stock price forecasts. Experimental results demonstrate that integrating sentiment analysis improves the model's predictive accuracy by reducing Root Mean Square Error (RMSE) and Mean Absolute Error (MAE) compared to standalone LSTM models. Additionally, a strong correlation is observed between public sentiment trends and stock price movements, confirming the significance of sentiment-aware forecasting in financial markets. This hybrid approach provides investors and financial analysts with a more comprehensive decision-making tool, allowing them to anticipate stock price fluctuations more effectively.