Research On The Study Evaluation Model Based On Bayesian Networks And Its Application In E-Learning System

This paper first analyzes the E-learning technology development status quo, pointing out that the existing deficiencies of E-learning system in the personalization,adaptability.Establishing a student model exactly reflecting student personal information is a crucial technology to provide individualized teaching.And it is essential to study assessment for building a student model. Only determining the exact knowledge level reached by the student at the different levels of granularity,the system can carry on the accurate judgment to student's general capacity,and provide the appropriate teaching strategy.So it is great significant to make use of computer technology and AI techniques to establish a reasonable and effective method to make study evaluation for individualized and adaptative teaching.In this paper,several student model assessment methods are compared and analyzed,and knowledge-based relations Bayesian Network study assessment model with the prediction capacity is put forward.By quantifying the uncertainty correlation between knowledge items,the study evaluation model has a strong prediction ability to be able to well reflect knowledge structure of students in the specific area.When study assessment model is set up,firstly,information of students will be divided into two parts.One is relating to the field knowledge,the other is not. The process of modeling the field knowledge information is the process of convert the course into Bayesian network.The paper makes "artificial intelligence" tutorial as an example,dividing the knowledge structure according to the knowledge items,then setting up the dependencies between knowledge items to determine Bayesian network causal inference relations,and finally determining a priori probability for each knowledge item.In conclusion,Bayesian network-based study assessment model is set up.In this paper,Junction Tree Algorithm is applied to inference and to update the model.Besides,through the simplification of network structure and optimization the order of deleting node in the process of triangulation algorithm, the time and space complexity is lowed.In this way,when information about the student changes,the model is able to update the probability distribution between knowledge items through Bayesian networks self-learning ability.Thus the ability and the study state of the student is predicted and reasoned.The basis for the personalization of teaching is provided.Finally,the authors developed an E-learning system based on Bayesian networks with the Microsoft Visual Studio 2005 platform and C # language,and made a simulation experiment.The experiment results show that the model can judge the mastering degree of students to the knowledge more properly and provide the students with individual leading strategies.