System Development Of Computerized Adaptive Testing

The exam is very widely used in social life, the traditional examination is the pen and paper test,which is gradually replaced by the computer-based test. But the ordinary computer examination is just a form changed in which all of the participants at the same time to answer the same questions, can not do "according to tester".This article designs an adaptive testing system,which can be set according to different participants, finally give the true ability of the participants. Computer adaptive test bases on item response theory,thus it firstly expounds the basic content and the advantages of item response theory, explains the specific meaning of each parameter in item response model.This article introduces the process of adaptive test, mainly divided into two stages:the primary ability estimation and the final ability estimation. Both stages include parameters estimated,selection algorithm, termination principle and so on. It mainly studys a new method in the primary of ability stage that the simulation test method,which involves a random number generation,Random walk algorithm and so on, finally through the contrast do have better ability of initial value.Establishment of adaptive test system, including the algorithm design and pro-gram implementation. System with three parameters model,Bayesian estimation,information and ability matching selection algorithm,the principle of variable length termination, based on database-Acess2000,used ourselves platform language to achieve, and which can be updated and management at any time.The end of the article is a summary and prospects future problems that need further study.