The construction of student-centered artificial intelligence online music learning platform based on deep learning

Area of Science:

• Educational Technology
• Artificial Intelligence
• Music Education

Background:

• Online music learning platforms require effective student-centered course recommendations.
• Existing models struggle to adapt to dynamic student learning interests and complex course interactions.

Purpose of the Study:

• To propose and evaluate a Course Recommendation Model for Student Learning Interest Evolution (CRM-SLIE).
• To enhance the accuracy and adaptability of course recommendations on online music learning platforms.

Main Methods:

• The CRM-SLIE model integrates an attention mechanism and Gated Recurrent Unit (GRU).
• It incorporates a project crossing module to capture evolving student interests and inter-course relationships.
• Model performance was evaluated across various embedding dimensions and student behavior sequence lengths.

Main Results:

• CRM-SLIE demonstrated excellent performance, achieving a highest Area Under the Curve (AUC) of 0.872 at an embedding dimension of 64 and sequence length of 20.
• The model achieved a maximum recall rate of 0.364, outperforming comparative models.
• Ablation experiments confirmed the significant impact of position coding and item crossing strategies.

Conclusions:

• The CRM-SLIE model exhibits strong adaptability, robustness, and practical value for online music course recommendation.
• It effectively captures dynamic student interests and complex course relationships, enabling personalized recommendations.
• The model significantly improves the learning experience by providing accurate and relevant course suggestions.