The Research And Design Of CAT System Based On IRT

Exam is one of the most important links in teaching activity,now,with the increasing maturity of the development of computer technology and network technology,many kinds of network examination system are emerging.Most of the test systems are based on the classical test theory(CTT) currently.Therefore,some problems exist such as tester's knowledge and ability levels can not be tested objectively and accurately and so on.In the thesis item response theory(IRT) will be used to solve the problems and achieve an adaptive test system which tests according to tester's aptitude.Computerized Adaptive Testing(CAT) is based on IRT,and it is the method of the computer technology applied in education measurement which has made significant progress in recent years.CAT makes up the deficiency of the traditional test which is based on CTT to a large extent.Practical CAT is rare in domestic,so the study of CAT system is space and practical value.In the thesis,the basic principle,model,parameter estimation and information function of IRT are discussed based on three education test theories.Thesis compares current three important algorithms of development CAT,.and a plan to realize CAT system is put forward.This systems use three parameters logistic model(3PLM) to describe the relationship between the project characteristic and tester's trait level.Then, create a test database based on the characteristics of 3PLM and CAT and discuss the estimation methods of parameters deeply.In the thesis,use the correlation between the static results of CTT and IRT to estimate project parameters,which is based on the invariance of IRT item parameter and with the advantages of CTT.At last,the author uses J2EE technology,three-tier MVC design pattern and B/S structure to achieve a CAT system which is based on IRT theory.The accuracy and efficiency of system is proved by some evaluation indexes through Monte Carlo simulation experiment.In the end,the thesis summarizes the full text and proposed the direction and content for further research.