摘要
该文依据项目反应理论(IRT)的0-1记分模式,应用BP神经网络对项目参数和考生能力进行估计。在对这些参数进行估计时,将得分矩阵分别采用得分率、通过率、相关系数、猜测率进行降维处理。此方法经计算机模拟实验,结果与目前流行的参数估计方法进行分析比较,能得到更好的精度,降低了误差。训练好的网络可以用于实际测试中,并且只需较少的人数和题数,能得到较好的精度。在计算机模拟实验过程中,考生能力和项目难度、项目区分度采用正态分布比均匀分布要好得多。
The item parameters and person ability in Dichotomously Scored of Response Theory model are estimated with Back-Propagation Neural Network.The dimension of Scoring matrixes X is descended by using scoring rate or passing rate or coefficient of correlation or guess rate when estimating those item parameters.The method is simulated in computer,and the results show that the item parameters estimation is more precise than the current prevailing software.The well -trained Neural Network can output the estimate value in field test and needs less number of examinees and items.The errors between estimate values and true values are very small.The experiment shows that the behavior of the simulation data drawn from normal distribution is better than those drawn from uniform distribution.
出处
《计算机工程与应用》
CSCD
北大核心
2004年第17期56-57,108,共3页
Computer Engineering and Applications
基金
国家自然科学基金项目(编号:60263005)
全国教育科学规划重点课题项目(编号:DBB010501)资助
关键词
IRT
参数估计
BP神经网络
Item Response Theory(IRT),item parameter estimation,Back-Propagation neural network
