• Thazin Aung

  Author

• Hnin Ei

  Author

This research proposes the design of an innovative knowledge graph-based personalized learning approach to resolve problems associated with the rigid structure of English learning processes and content-oriented recommendation systems. This approach tries to integrate learner profiles, English subject connections, learning analytics, and adaptive recommendations in order to deliver continuous personalized learning paths. The framework consists of data acquisition, knowledge graph creation, learning analytics, and recommendation components. The data acquisition component includes information about studied topics, difficulty, preferences, test scores, task completion, and time spent. Knowledge of vocabulary, grammar, and reading subjects is defined as interrelated knowledge graph nodes. Learning analytics identifies learners' strong and weak aspects and recommends personalized learning activities accordingly. Classroom observations and before/after percentage comparisons were applied for the assessment of learning outcomes and teacher and learner feedbacks. Overall average learner's results improved from 48.4% prior to implementation to 81.0% after implementation, resulting in a 32.6 percentage point improvement. Vocabulary building improved from 48% to 82%, grammar understanding from 56% to 85%, reading confidence from 45% to 79%, writing practice motivation from 38% to 72%, and engagement from 55% to 87%. Average teacher and learner feedback increased from 51.0% to 85.3%, gaining 34.3 percentage points. The learner satisfaction score achieved 88%, and the recommendation relevancy had the highest increase in the feedback rate at 38 percentage points. The results suggest that combining knowledge graphs and learner analytics is capable of creating an interconnected, adaptive, and engaging English language learning environment without making teachers go through much effort in choosing lessons. The proposed model appears to be a potential supplement to rigid teaching, even though more research is required to verify its effectiveness.

• PDF