摘要：Cognitive diagnostic assessment (CDA) can get some information about the student's cognitive processes and knowledge structure based on the psychometric model. Most of the previous studies using traditional cognitive diagnosis models (CDMs) based on the cross-section. This study aims to propose a new longitudinal cognitive diagnosis model, namely the HMM/ANN model. In this model, the artificial neural network (ANN) was used as the measurement model of the hidden markov model (HMM) to realize longitudinal tracking of students' cognitive skills. This study includes simulation study and empirical study. The results show that the HMM/ANN model can confirm high classification accuracy and correct conversion rate when the number of attributes is small. The combination of ANN and HMM can help to effectively track the development of students' cognitive skills in real educational situations. Moreover, the classification accuracy of HMM/ANN model is affected by the quality of items, the number of items, and the number of attributes examined, but not by sample size. The classification result and the correct transition probability of the HMM/ANN model were increased with the increase of the item quality and the number of items but decreased with the increase of the number of attributes.
