Recurrent Neural Cognitive Diagnosis with Semantic Analysis and Self-attention

With the continuous development of educational technology, cognitive diagnosis is essential in intelligent education. This study proposes a cognitive diagnosis model called RNCD(Recurrent Neural Cognitive Diagnosis), which constructs a comprehensive feature vector by integrating student embedding, exercise discrimination, and knowledge relevancy text embedding. The main goal is to utilize Recurrent Neural Networks (RNN) combined with a self-attention mechanism and semantic analysis. This approach aims to achieve the prediction of students' proficiency in different knowledge points. Based on the feature construction, we use RNN layers for sequence modeling to capture students' learning history on each knowledge point effectively. At the same time, a self-attention mechanism is introduced, which enables the model to learn and weigh important features automatically, thus improving the model's representative ability and predictive performance. Regarding semantic analysis, the knowledge relevancy vectors are converted into text embedding vectors by a linear transformation to capture the correlation between knowledge points comprehensively. The comprehensive features of the RNCD model combine recurrent neural networks, self-attention mechanisms, and semantic analysis, which can capture students' cognitive states more effectively and accurately.