Simulation of personalized english learning path recommendation system based on knowledge graph and deep...

Area of Science:

• Educational Technology
• Artificial Intelligence
• Computer Science

Background:

• Online education is rapidly growing, increasing the need for effective personalized learning path recommendations.
• Current recommendation systems struggle with knowledge structure modeling, dynamic learner state tracking, and optimized strategies.

Purpose of the Study:

• To propose an advanced online personalized English learning path recommendation method.
• To address limitations in existing knowledge structure modeling, learner state perception, and recommendation strategies.

Main Methods:

• Integrated a domain knowledge graph with deep reinforcement learning (DRL).
• Modeled prerequisite and semantic relations, mapping resources to concepts.
• Updated learner mastery using real-time feedback, graph propagation, and forgetting mechanisms.
• Formulated the task as a Markov Decision Process (MDP) using Q-learning and Proximal Policy Optimization (PPO).

Main Results:

• Achieved high performance metrics: Precision 0.85, Recall 0.82, F1 0.84, MAE 0.12, RMSE 0.18.
• Demonstrated superior performance against multiple strong baselines.
• Showcased practical scalability with low latency (241 ms) and fast startup times.

Conclusions:

• The proposed method effectively enhances personalized English learning path recommendations.
• The integration of knowledge graphs and DRL offers a robust solution for online education.
• The system is scalable and suitable for real-time application in educational platforms.