Researches Of Across-Discipline Intelligent Tutoring System Based On Domain-Specific Ontology

National Knowledge Infrastructure (NKI for short) is being built to provide an effective knowledge service for people in general and a knowledge foundation for hi-tech products in particular. The NKI will contain knowledge from the Chinese encyclopedia consisting of 71 disciplines within 74 volumes. How to design and implement a tutoring system to help people to acquire knowledge is an important research area on application.We are building an Intelligent Tutoring System called NKI-Tutor to help people to acquire the abundant knowledge from NKI.There have been a lot of work on Intelligent tutoring systems,and they are viewed as cost-effective alternatives to traditional education. However,there are some limitations in the traditional systems:First,the teaching material is from specific domain knowledge,and the teaching strategy can rarely teach the across-discipline.Second,conventional ITS concentrates more on advancing a student's state of knowledge than analyzing and improving the student's cognitive state and learning style etc.Third,structure of conventional ITS rarely include simulation tools and Exercise Module.This thesis contributes in the following aspects:Building an ITS system model called NKI-Tutor. The conventional systems contain 5 aspects:Student Model,Pedagogical Module,Domain Knowledge,Communication Module and Expert Model. However,the system ought to offer some simulation tools to help users to realize the procedural knowledge and strategic knowledge in the Interactive Learning Environment. At the same time,Exercise Module is a very important model.Building an estimate method for the state of student based on an abstract of Domain-Specific Ontology. The component of student model is an essential part of an ITS. So far there are lots of models and theories in the field,which,however,concentrate more on advancing a student's state of knowledge than analyzing and improving the student's cognitive state. A compact knowledge model,called Concept-Relation Model (or CR model for short),is used as a level of abstraction over different disciplines. The method includes three aspests:abstract of Domain-Specific Ontology,cognitive ability and estimate rules,andwhich explicitly utilizes both tests and the domain conceptual structure in ability evaluation.Building a teaching strategy for Across-Discipline based on an abstract of Domain-Specific Ontology. NKI covers the Domain-Specific Ontology on the knowledge of different disciplines,the teaching strategy in terms of the abstract of the Domain-Specific Ontology makes Across-Discipline teaching come true,and also makes NKI-Tutor teaching individualized. We design the teaching strategy from two level:teach attributes and relations (call for micro teaching strategy),teach a concept or a curriculum (call for macro teaching strategy) et al.Designing an exercise module. One of the labor-intensive tasks in developing an intelligent tutoring system is to manually design tests to determine,posterior to a single lesson,whether a student has properly mastered the taught concepts. We design a framework for automation tests. The core of our framework consists of three components:A conceptual model for a domain to be taught to students,A collection of testing goals to indicate what are needed to be tested,and A collection of test-generation rules for achieving those goals.