| In this dissertation, Artificial Teaching Assistant (ATA), a framework for intelligent tutoring systems (ITS) has been proposed; further, the design and implementation of it have been described. The promise of the framework for adaptive tutoring is based on both a unified understanding of knowledge, learning, instructional design, and learners' motivations and a unique use of a hybrid knowledge representation model with the rough set user model built on top of it.; On the conceptual level, the ATA framework consists of three interrelated models--presentation, question, and test--which are central aspects of a tutoring process and suggest the importance of both the organization of knowledge to an automatic and effective presentation and the provision of questions and tests to the assurance of students' learning of presented knowledge.; On the design level, the architecture underlying the ATA framework consists of two components: knowledge representation modeling and user modeling components that are designed to capture two types of knowledge possessed by a tutoring system. The two types of knowledge are the subject material to be presented and the tutoring knowledge needed to make an effective presentation based on the various responses of the student during tutoring.; The purpose of designing such ATA tutors has been to assist the teaching of scientific knowledge. The primary goal of the tutoring system is to facilitate mandatory learning. It aims to relieve the laborious portion of the instructor's duty in teaching by building an autonomous tutoring agent to teach students in whatever subject they are supposed to master.; The novelty of this computational approach to effective tutoring is grounded on the constructivist learning of abstract and mathematical concepts that are acquired by applying constructivist procedures to a basis of finite lexical concepts expressed in natural language. The unique classification of knowledge, especially abstract knowledge, enables the accessing of knowledge in the system precisely and automatically. Thus, the mechanistic knowledge representation model (the G-S model) combined with the approximate reasoning mechanism (the rough set model) may be applied to achieve autonomous tutoring that should be both effective and adaptive. Further, the implementation and evaluation of a prototype system, number systems tutor (NST), has demonstrated the possible reusability of components. |
