The neural cognitive diagnostic model integrated item weight based on information function(WNCD)is proposed,which adjusts the weights of test items for inferring student cognitive status based on the proximity esti-mated between student abilities and item difficulties.The model also utilizes neural networks to capture the complex nonlinear interactions between test items and students.Comparison experiments with existing methods on ASSIST and PISA2012 data sets show that WNCD not only maintains good interpretability,but also outperforms the equal-weight neural cognitive diagnosis models on diagnostic accuracy.
关键词
认知诊断/神经网络/试题权重/信息函数
Key words
cognitive diagnosis/neural network/item weight/information function
维普
万方数据