用于项目生成的认知模型的构建与比较——以矩阵完成问题为例

The Construction and Compares of Cognitive Models for Item Generation——Taking the Matrix Completion Problems for Example

项目生成是一种新的测验编制技术,它可以弥补传统测验编制技术的缺陷。使用该技术编制测验,要进行大量的前期工作,如必须要了解和归纳所编测验中项目的所有刺激特征,据此建立认知模型,再将认知模型与心理计量模型联合,构建能预测新生成项目难度的数学模型等。该研究以矩阵完成问题为例,在带约束的两参数Logistic模型的基础上,通过对构建的几个认知模型的比较,挑选合适的认知模型为矩阵完成问题的项目生成研究服务。研究结果表明,自建的认知模型能够满足矩阵问题项目生成的要求。

Item generation is a new test construction technology,which can make up for the technical defects of the traditional test construction. To use this technology to construct test,it is necessary to do a lot of preparatory work,such as the need to understand and summarize all stimulus characteristics of the items in test,to establish cognitive models accordingly,then to building mathematical models which can predict the difficulty of each generated item by unifying cognitive models with psychometric ones,and so on. Taking the Matrix Completion Problems for example,the research would serve to generate items,by comparing cognitive models to select a suitable one,on the basis of the 2PL - constrained model. The results showed that the self - built cognitive model could meet the requirements of the item generation of Matrix Completion Problems.