Leveraging Deep Learning for Social Media Behavior Analysis to Enhance Personalized Learning Experience in Higher Education: A Case Study of Computer Science Students

Abstract

This study investigates the application of deep learning techniques for analyzing social media behavioral data to enhance personalized learning experiences in higher education, specifically focusing on computer science students. The research implements a sophisticated deep learning framework incorporating LSTM networks and attention mechanisms to process multi-modal social media data streams and predict student performance patterns. The methodology encompasses the collection and analysis of social media interaction data from 1,245 computer science students across multiple platforms, employing advanced feature engineering techniques for behavioral pattern extraction. The developed model achieved 93.8% accuracy in predicting student performance trajectories, representing a 15.3% improvement over traditional methods. Analysis revealed significant correlations between specific social media engagement patterns and academic outcomes, with high-interaction students demonstrating 24.3% better performance compared to minimal-engagement groups. The framework successfully identified at-risk students with 89.2% accuracy within the first four weeks of the semester, enabling proactive intervention strategies. This research contributes to both theoretical understanding of digital learning behaviors and practical implementation of personalized learning systems in higher education, establishing a novel paradigm for integrating social media analytics with educational technology.