One-on-One Tutoring by Humans and Computers articulates the CIRCSIM-Tutor project, an attempt to develop a computer tutor that generates a natural language dialogue with a student. Editors Martha Evens and Joel Michael present the educational context within which the project was launched, as well as research into tutoring, the process of implementation of CIRCSIM-Tutor, and the results of using CIRCSIM-Tutor in the classroom. The domain of this project is cardiovascular physiology, specifically targeting first-year medical students, though the idea is applicable to the development of intelligent tutoring systems across populations, disciplines, and domains.
This 5 year-long project was motivated by the belief that students need assistance in building appropriate mental models of complex physiological phenomena, as well as practice in expressing these ideas in their own words to fully develop those models, and experience in problem-solving to use those models effectively. The book outlines directions for future research, and includes distinct features such as:
*detailed studies of human one-on-one tutoring;
*learning outcomes resulting from use of the tutor;
*natural language input parsed and translated into logical form; and
*natural language output generated using the LFG paradigm.
This volume will appeal to educators who want to improve human tutoring or use computer tutors in the classroom, and it will interest computer scientists who want to build those computer tutors, as well as anyone who believes that language is central to teaching and learning.
Table of Contents
Contents: Preface. Part I: Introduction. Origins of the CIRCSIM-Tutor Project. What's Being Tutored: The Domain and the Problem. Part II: Studies of Expert Human Tutors. The Collection and Analysis of Tutoring Transcripts. The Language Used by Tutors and Students. Domain Knowledge for Problem Solving and for Tutoring. Planning for Tutoring: What Is the Task to Be Accomplished? Carrying Out the Tutoring Task. Hinting as a Technique to Keep Students Active. A Tutoring Episode Analyzed. How Effective Are Human Tutors? Part III: Building CIRCSIM-Tutor: An Intelligent Tutoring System With Natural Language Capabilities. The Architecture of an Intelligent Tutoring System. Plans and Planning in the CIRCSIM-Tutor Project. Understanding the Student Input. Student Modeling. The Domain Knowledge Base and the Problem Solver. Generating Tutorial Dialogue. The Screen Manager. Part IV: What Have We Learned? CIRCSIM-Tutor in the Classroom. Machine Tutors and Natural Language. Where Do We All Go From Here? What We Have Learned.
"...recommend this book to all researchers who are interested in explaining problem oriented human-human dialogues or who are interested in building and evaluating dialogue systems as one way of testing the generality of one's theories and systems....a guidebook for researchers who are interested in getting started in the area of dialogues for educational applications."