一种非参数的多策略方法:多策略的海明距离判别法

The Formulation of Multiple-Strategies Hamming Distance Discrimination

将单策略汉明距离判别法(SS-HDD)拓展至多策略情境(MS-HDD),并通过多因素的模拟设计及与已有研究的比较,考察其精确性与适宜性.研究结果表明:1)MS-HDD构建思路简洁,无需前提假设和参数估计,计算简便;2)单策略情境:MS-HDD与SS-HDD具有等同的判别效果,且判准率均高于相同条件下的广义距离模型(GDD)和多策略广义距离模型(MSCD);3)多策略情境:MS-HDD具有与MSCD相当的PMR和SMR,MS-HDD可适于多策略测验情境.

The purpose of the paper is to formulate a nonparametric multiple-strategies cognitive diagnosis discrimi-nation method, which called Multiple-Strategies Hamming Distance Discrimination （ MS-HDD） , to estimate exami- nees7 knowledge states that generated in multiple-strategies of problem solving. The results show that under the sin-gle-strategy testing situation,all of the SMR,AAMR and the PMR of MS-HDD method are as similar as SS-HDD^s results. Under the multiple-strategies testing situation, the MS-HDD leads to more accuracy in these indexes than the SS-HDD. Under the multiple-strategies testing situation,MS-HDD is compared with MSCD,and the result show that MS-HDD has more better performance than MSCD when the random slip is growing. All in all, the method that ex-tends the SS-HDD method to all for multiple-strategies of problem solving,which is called MS-HDD,was suited to multiple-strategies testing situation and could be accepted in our practical application.