Research On Bayesian Network Knowledge Reasoning In ITS Based On CBR

Intelligent Tutoring System,as an important research field of computer network technology and artificial intelligence,applying artificial intelligence in education research and practice,heads the 21st digitalize education of human society.Traditionally,ITS is comprised of three main models:knowledge base(field knowledge and experiment knowledge),student model(students' knowledge level and study ability),teaching model(teaching strategies) and human-machine interface(friendly communication among ITS,teachers and students ).Knowledge base,as an important constitution of ITS,which means the ITS should recommend appropriate teaching content and teaching strategies to students according to their study interests,study preferences, study style and knowledge background.Thus,knowledge represent and knowledge reasoning are the bottlenecks of ITS.In order to facilitate knowledge represent and knowledge reasoning,the paper managed to apply Corpus similarity measurement and Bayesian knowledge based on students' study preferences to knowledge base model of ITS,implement teaching in accordance of students' aptitude and collaborative studying among students.This paper proposes a knowledge reasoning network and a knowledge reasoning algorithm based on Bayesian network,which derives and recommends the suitable learning resources and teaching methods from the existing learning resources pool and teaching methods base to students.Thus,the intelligent recommendation function of ITS is achieved.Experiments shown:This method can achieve excellent recommendation effect,but the searching space is comparatively large. Meanwhile,two problems exist:(1) Fragmented the relationship of knowledge reasoning module and evaluation systems,choosing the appropriate teaching strategies and the next system behavior from the teaching strategies base without considering the evaluation of students' feedback.(2)Teaching strategies can not be shifted with good flexibility according to changes of students' situation,so certain limitations exist. Therefore,the Bayesian network inference based on CBR(Case-based Reasoning) is proposed in this paper,constructed a suitable description of cases using Bayesian network,assessed and retrieved the cases through technology of probability spreading,achieved machine learning through revising the acquired case and updating the conditional probability of Bayesian networks.Achieved the reuse of teaching experience and implemented collaborative learning by adopting multi-mode dynamic teaching strategies model and employing a similar teaching strategy to similar students.The experimental results show that the algorithm greatly enhanced the performance of the recommendation system;meanwhile,the searching space is greatly reduced.