Research On Cognitive Diagnosis Methods Based On Knowledge Attributes

In the Educational informatization 2.0,China clearly points out that it is necessary to promote the development of the new form of "Internet+Education" and construct education system,which is intellectual,individual and lifelong to fulfill student’s needs on personalized learning and promote the sustainable development of education.Personalized learning material recommendation,planning learning path and teaching guidance have become a significant topic in recent education domain.Knowledge attribute,that is,the knowledge points or skills examined in the test items.Cognitive diagnosis theory bases on student’s question answering behavior to diagnose student’s knowledge concept mastering level.It can be used to reveal student’s advantage and disadvantage which could help to provide student with personalized learning path and intervention.So,the research on cognitive diagnosis theory is of great significance.All our research works are based on this theory,and the main points are as follows;Firstly,we investigated history of cognitive diagnosis model’s development.We gave a detailed description of frequently used Q-matrix cognitive diagnosis models and analyzed their advantage and disadvantage.The models mentioned here includes the rule space model(RSM),Deterministic Inputs,Noisy "And" gate model(DINA)and uniform model(UM).Secondly,in order to improve the effect of incomplete knowledge attribute description in Q matrix on cognitive diagnosis,the concept of supplementary matrix of Q was innovatively proposed to improve the cognitive diagnosis process of basic DINA model.and a cognitive diagnosis model based on incomplete Q matrix was proposed.Experimental results show that the cognitive diagnosis model based on incomplete Q matrix can effectively improve the accuracy of cognitive diagnosis compared with the basic DINA model.Thirdly,aiming at the influence of knowledge attribute hierarchy on students’cognitive process,a cognitive diagnosis method based on DeepFM model was proposed to learn the complex knowledge attribute hierarchy based on DINA model.Through experimental comparison and analysis,the results of the cognitive diagnosis method based on the DeepFM model can improve the prediction accuracy of students’ answer results,which proves that the cognitive diagnosis method based on the DeepFM model can improve the diagnosis effect of students’ cognitive state.Finally,in this paper,based on the knowledge attributes,a cognitive diagnosis method based on the incomplete Q matrix and a cognitive diagnosis method based on the DeepFM model are proposed by exploring the common problems,that is,the influence of the incompleteness of Q matrix and the hierarchical relationship of knowledge attributes on the results of cognitive diagnosis.In order to explore the performance of the two methods,the two methods were verified experimentally on public and private data sets respectively,and compared with other models.The results show that both the cognitive diagnosis method based on incomplete Q matrix and the cognitive diagnosis method based on DeepFM model can significantly reduce the diagnostic errors of the cognitive diagnosis model and improve the diagnostic accuracy of students’ capacity.