Research And Implementation Of Learning Diagnosis Model Based On Knowledge Space Theory

In the education domain, how to diagnose and test the situation of mastering the knowledge of the students accurately and efficiently is a hot spot of study. The modern test theory demands that test report should not only give a test or ability scores, but also to provide the knowledge structure, cognitive defects and other information. Knowledge space theory, which provides a way to express the knowledge structure, can be used to diagnose and test the level of knowledge and knowledge structure of the students and has developed into the most effective knowledge representation language in adaptive teaching and testing system. This article focuses on the design and implementation of diagnostic model based on knowledge space theory. Its main tasks are the following aspects.First, an adaptive diagnosis algorithm based on the relationship tree of questions was proposed by analyzing the premise relationship between questions when the questions contain more than one aggregate of knowledge points. Considering a question probably comprises a variety of knowledge point aggregates for existing different problem-solving methods in the actual test process, discussed the possible three types of relationship between questions: AND relationship, OR relationship and independent relationship. Based on AND relationship and OR relationship, the relationship tree of questions formed through the tree structure of questions. Designed and implement the algorithm used to construct the relationship tree of questions in the paper. Then proposed a diagnosis algorithm based on the relationship tree of questions, the algorithm starts from the root node of the tree according to the breadth priority rule, it could speculate the answer case of sub-nodes based on the answer case of parent node, the questions which could be speculated the answer state would not be tested. Finally, an algorithm used to generate the diagnosis decision tree was given. The diagnosis algorithm wouldn't need to establish knowledge space, could save the test time and applied to acquire the students'mastery situation at each knowledge point after the chapter study.Second, a learning diagnosis model based on the guess probability and misplay probability was proposed, the model considered the possibility of guess and misplay in actual testing process. It believed that the students may guess the answer or misplay the answer, so the answer state of the students may not the real reaction. The concepts of guess probability and misplay probability were proposed, the calculation method of them was discussed in detail. In a group of questions, when the answer state of students conflict with the premise relationship, there must be exist the possibility of guess and misplay. According to the probability of guess and misplay of each question, the model could transfer the antipathic answer state to some consistent answer state with the largest probability, return the corresponding diagnostic results, the process is called the answer state transfer. The diagnostic feedback information of the model includes not only the knowledge structure of the students but also the guess probability and misplay information that may occur in a testing process.Third, designed and implement a study diagnostic system. The diagnosis system has two modes: the adaptive diagnostic model and the probability diagnosis model, which implement the diagnostic test algorithm based on the relationship tree of the questions and the diagnosis model based on the guess probability and misplay probability separately. Introduced the work flow and database design of the system, took the quadratic equation with one unknown in math course of middle school as an example and demonstrated the study diagnosis process under the two models.