Intelligent Sentiment Prediction in Social Networks leveraging Big Data Analytics with Deep Learning

Cloud computing has recently made it easier to distribute varied, unstructured digital data within social networks of differing opinions. Processing large volumes of text data requires precise computational methods, which increases the system?s workload. Integrating big data with Natural Language Processing (NLP) has enhanced this. Frameworks like MapReduce enable parallel computation for large tasks. This study proposes a smart sentiment model based on Deep Learning (DL) and batch and streaming big data analytics. The research aim is to utilize the power of distributed platforms for real-time data processing. These platforms assist with data cleaning, size reduction, decreasing access times, and reducing storage needs. This preprocessing step makes the streaming more suitable for data-intensive models. This research focuses on handling large-scale, short-text data using batch and streaming frameworks combined with DL techniques in NLP. We present a method to analyze short texts, determine their semantic meaning, and classify them into positive or negative sentiments. The process involves data reduction and refinement using selected features and big data tools, followed by embedding words with global vectors to feed into convolutional and RNNs. The experimental results confirm the effectiveness of our approach, demonstrating its superiority over existing methods. Our model achieved a remarkable accuracy of 97.31%.