Research On Selection Strategy Based On Item Response Theory Of Computer Adaptive Testing

Item response theory is one of the modern psychology and education measure theory, itsmain application is a computer adaptive test. At present, computer adaptive test is becomingmore and more,it is applied to education exam. From the school point of view, computeradaptive test can assist teaching test, reduce the time of the teachers grade examination papers.From the social perspective, adaptive test system can reduce test staff batch marking sheet oftime, it also reduces the material resources and manpower, it makes for test results statistics.From test teaching point of view, this papers establishes adaptive test system, it can providebetter service for teachers and students, and provides convenience for teaching and help.At first, this paper analyzes in detail item response theory concept, basic assumption,characteristics, content and model, and then analyzes computer adaptive test principle,realization conditions, implementation process, the termination conditions, etc, and finally usesJava as a development language, Tomcat6.0as a Web server, MySQL5.0as the database server,using the Eclipse integrated development environment, design and implementation based onWeb computer adaptive test system, the system mainly include user management, result inquiry,test management, benefits management, question bank management function module. Thisarticle main research the topic during the implementation of adaptive testing strategy, usingthree kinds of topic selection strategy, the maximum information method, the maximumdifficulty method, two methods combined. the maximum information method is currentlyselection strategy method, the maximum difficulty method and two methods is in this paper. Inadaptive test system, it designs experiment purpose, experiment contents and experimentprocess, through the experimental data, the research and analysis of maximum difficultymethod, two method combined is effective, and shows adaptive test system is efficient, andmake evaluation on the performance of the system.Based on the experiment proved that maximum difficulty method, maximum difficultymethod and two method combined is effective. Through experimental comparison and analysis,two method combined is better test efficiency, test error is small, test standard error is minimum,it is proposed in this paper a close with maximum information methods a new selectionstrategy.