摘要
提出了基于神经网络的PSP方法并将其运用于认知诊断,以488名高三学生参加语文模考的数据为例进行分析,结果表明:(1)主成分分析方法可以对SOM网络的输入数据进行降维,并得到测验涉及的认知成分。(2)SOM网络可以对被试进行分类,并由此得到各类的认知缺陷。(3)概率神经网络可以对新的被试进行较为准确的类别判断。PSP方法是一种较为实用的认知诊断方法。
The neural network-based PSP method was put forward and applied in cognitive diagnosis. The data obtained from 488 students of high school grade 3 who sat in a Chinese simulation test were analyzed. The results showed that （ 1 ） principle component analysis could reduce the dimensions for SOM input data, and get the cognitive attributes; （2）SOM network could divide the subjects into clusters, and get the cognitive shortages of different categories; （3）probabilistic neural network could judge the new subjects accurately. PSP was an applied cognitive diagnostic method.
出处
《心理科学》
CSSCI
CSCD
北大核心
2010年第4期915-917,共3页
Journal of Psychological Science
基金
国家社会科学基金"十一五"规划课题(BBA080050)
江苏省教育科学"十一五"规划课题(D/2008/01/105)
关键词
认知诊断
神经网络
主成分分析
SOM网络
PNN网络
cognitive diagnostic, neural networks, principle component analysis, SOM network, probabilistic neural network
