Scalable vortex search-tuned intelligent adaptive boosting for sentiment analysis of social media data using natural language processing

• Deals with imbalanced and noisy social media text with strength by using TF-IDF and Z-score normalization. • Scalable real-time sentiment analysis through integrating Scalable Vortex Search-tuned Intelligent Adaptive Boosting (SVS-IAda Boost). • Facilitates real use, like crisis tracking and brand sentiment measurement. • Improved on conventional ensemble classifiers by 1.7% to 4.7% points on sentiment prediction. • Obtained the accuracy of 95.1%, recall of 91.6%, and 92% F1-score on the true Twitter data. Sentiment analysis serves an important role in comprehending public perception, behaviour, and emotional reactions on social media platforms. With the growing influence of online content in shaping societal attitudes, analysing sentiments in real-time has become vital for sectors like healthcare, politics, and marketing. Despite significant advancements in Natural Language Processing (NLP) and machine learning, challenges remain in accurately classifying sentiment from short, noisy, and often imbalanced social media data. The research addresses these limitations by proposing a novel hybrid model, Scalable Vortex Search-tuned Intelligent Adaptive Boosting (SVS-IAda Boost), designed for enhancing sentiment classification performance. The model was tested on a large-scale Twitter dataset (more than one million posts) with a varied range of real-world events to determine whether the model would recognize those events as positive, neutral, or negative. For feature extraction, the Term Frequency-Inverse Document Frequency (TF-IDF) technique is employed, optimized for sparse text data. The proposed SVS-IAdaBoost framework integrates the exploratory capabilities of the Vortex Search algorithm with the adaptive ensemble learning of IAdaBoost to boost classification accuracy and handle imbalanced sentiment distributions. The SVS-IAdaBoost model outperformed conventional classifiers in terms of accuracy (95.1%), recall (91.6%), precision (93.4%), and F1-score (92.0%) across all sentiment categories. This research demonstrates the viability of scalable hybrid models for real-time sentiment analysis and contributes to the development of more responsive and intelligent decision-support systems based on social media analytics.