A HYBRID APPROACH OF CLUSTERING AND SWARM INTELLIGENCE FOR EFFICIENT SOCIAL MEDIA TREND EXTRACTION IN LARGE-SCALE DATA ENVIRONMENTS

Abstract

Twitter, Facebook, and Instagram, among other social media sites, produce enormous amounts of unstructured data in every passing second in the modern digital age, posing daunting challenges for extracting real-time meaningful and actionable trends. Conventional analysis techniques are typically unable to handle the large velocity, high volume, and wide variety of social media information, hindering their capability in large-scale settings. This study suggests a hybrid model that combines clustering algorithms with swarm intelligence approaches in order to effectively extract breaking trends from huge social media datasets. The suggested methodology starts by using unsupervised clustering (k-means clustering) in order to cluster similar social media posts based on textual content, leveraging TF-IDF vectorization in order to represent the data. In order to address the shortcomings of cluster algorithms—such as being sensitive to initial centroids and unable to handle noise well—Particle Swarm Optimization (PSO), a well-known swarm intelligence algorithm, is used to optimize cluster centroids. The hybrid approach improves cluster quality by reducing intra-cluster distances and maximizing inter-cluster distances to enhance overall trend consistency. Experimental verification was implemented on big Twitter datasets with millions of public tweets. Experiments indicated that the hybrid method had much better accuracy in trend detection, improved processing speed, and improved scalability than the older clustering-only models. The model performed well in handling noisy and sparsely presented data, and adaptive optimization enhanced the usefulness of the extracted trends. These results demonstrate the promise of hybrid intelligent systems to meet big data problems in social media analytics. The methodology can be used in real-time trend identification, providing great advantages in market research, public opinion analysis, and decision-making in ever-changing digital landscapes.