Research on Teaching Resource Recommendation Algorithm Based on Deep Learning and Cognitive Diagnosis.

Area of Science:

• Educational Technology
• Artificial Intelligence
• Cognitive Science

Background:

• Networked educational resources are abundant, but current recommendation algorithms often fail to meet individual student needs.
• Existing systems inadequately consider student personality and cognitive characteristics, leading to generic recommendations.
• This gap hinders personalized learning and optimal student performance.

Purpose of the Study:

• To develop an improved teaching resource recommendation method that addresses limitations of existing algorithms.
• To enhance personalized learning by integrating cognitive diagnosis and student historical data.
• To boost student learning efficiency and academic performance through individualized resource suggestions.

Main Methods:

• Developed a cognitive diagnosis model using the Two-Dimensional Information-Need Analysis (TDINA) model.
• Proposed a Convolutional Neural Network (CNN) and joint probability matrix decomposition (CUPMF) based recommendation method.
• Integrated student answer history, cognitive abilities, knowledge mastery, and forgetting effect factors into the model.

Main Results:

• The CUPMF model effectively predicts student performance on educational resources by leveraging CNN for feature extraction.
• The TDINA model provides a knowledge mastery matrix for personalized recommendations.
• The integrated approach successfully recommends tailored teaching resources, enhancing learning outcomes.

Conclusions:

• The proposed recommendation system effectively addresses the limitations of traditional methods by incorporating cognitive and historical data.
• This approach significantly improves the personalization of educational resource recommendations.
• The study demonstrates a viable method for enhancing student learning efficiency and performance in digital learning environments.