Research And Implementation Of Adaptive Testing Method Based On Knowledge Space Theory

With the rapid development of the computer science technology, the application of computer network is increasingly, and then testing use of computer come to the world, which take use of the advantage of computer. It's not only the viewer of conventional test but also a testing tool, having features of fairer, safer and more efficient. Knowledge Space Theory can provide both the knowledge structure and the cognitive defects of the student, accelerating learning efficiency in subsequent study. As a result, we choose Knowledge Space Theory.It is not easy for teachers or experts to build a well knowledge structure according to the relationship between questions. In further study we find that knowledge structure based on knowledge points is very helpful for computer adaptive testing. So, we extended the present Knowledge Space Theory based on questions to a new Knowledge Space Theory based on knowledge points, and established a knowledge-point-based knowledge space, in which knowledge points is the basic element, and then the and relationship and or relationship between questions were proposed according to the ancestor relationship, rely relationship, brother relationship and parallel relationship between knowledge points. In the end, the relationship tree of questions was build, and it is founded of a well graded knowledge structure used in adaptive test.On the other hand, at present, most of selection strategy in computer adaptive test based on Knowledge Space Theory is dichotomous method. This method has a rigorous mathematical logic, but has less suitable. Person is in on a test, there are many uncertain factors that will affect the testing process. In this dissertation, neighbors of knowledge states were brought in adaptive test, and proposed a new selection strategy. This method makes adaptive test to fit for knowledge features of the student, and it can provide the knowledge structure and the cognitive defects of the student faster with less questions. It can better reflected the adaptive test process in the "varies".The final destination of test is to obtain some kind of abilities and the cognitive defects of the student, but we only get a knowledge state for reference at the end of test. The knowledge state is composed of some certain questions that have no common meaning. In this dissertation, we use of the relation of knowledge points and skills, give the method to build skill function, question function and the conversion between them. Furthermore, we also define skill state function that can connect knowledge structure and skill structure used in adaptive test. That is to say, we complete the mapping of knowledge structure onto skill structure, and in the end we can get the skill level of the student.