Tipping points in educational engagement: Sentiment analysis of online learners’ feedback

Massive Open Online Courses (MOOCs) represent a transformative shift in global education, yet learner satisfaction and engagement remain precarious—often subject to tipping points caused by usability flaws, content mismatches, or insufficient interactivity. This study investigates such critical thresholds in the “Introduction to Cyber Security” MOOC offered via India’s SWAYAM platform by analysing five years of learner feedback using AI-driven sentiment analysis and Topic Modeling. Drawing on 468 qualitative reviews from Class Central, our analysis identifies content relevance, technical reliability, and interactivity as key factors correlated with learner sentiment, with recurring technical issues and limited interactivity posing potential tipping points that could trigger sudden declines in learner engagement. While 73.5% of responses are positive, neutral and negative feedback clusters around system instability, lack of advanced content, and limited peer engagement—revealing vulnerability to disengagement. This work proposes a framework for real-time, adaptive course improvement using accessible NLP tools and highlights the potential for early warning indicators of learner dissatisfaction. The findings offer actionable guidance for MOOC designers, platform developers, and policymakers to sustainably enhance quality and resilience in digital learning ecosystems through AI-based feedback systems.