结合α分层改进基于IRT计算机自适应题库系统

Improvement of IRT computer adaptive item bank system based on the combination ofα stratification

IRT理论中基于最大信息量函数的选题方法会使系统偏向于选取信息函数值最大的试题进行测验,容易造成部分试题被反复选取,导致试题的曝光度过高从而影响后期测验的有效性。本文在原有选题策略的基础上融入α分层和内容筛选,有效地提高了系统的有效性和安全性。

The method of the maximum amount of information function based on IRT theory makes the system tend to choose the maximum value of test information function test, causing part of the test was repeated selection easily, resulting in a high visibility of test, thus affecting the effectiveness of post test. This paper combinesαstratification and content filtering based on the original selection strategies , effectively improving the effectiveness and safety of the system.