Course-oriented Knowledge State Model

Knowledge state indicates the level of mastery of a learner for each knowledge point in a domain, and Knowledge state model uses learners' performance in the filed of education to predict their future knowledge levels, is a well established problem in computer supported education. In this paper, we use course performance data to track the knowledge state of college students' academic process from the perspective of professional competency requirements. Course information reflects a more comprehensive professional competence of learners, and the use of course performance data with a value range of 0 to 100 makes the model more accurate compared to the binary variables of traditional answer sequences. Firstly classifying courses using hierarchical clustering combined with box-line plots, courses are divided into five categories, and explain the category results with practical implications; Secondly combining learners' future performance on courses with their past responses using an attention mechanism, and visualize the attention weights of knowledge sequences; Finally inputting neural networks to predict students' performance in each category of courses and outputting students' personalized knowledge state visualization chart. The experiments show that the proposed model is able to predict the learner's level of performance more accurately, which is potential for personalization in real-world educational settings.