397 Instructional Science 12 (1983) 397 398 Elsevier Science Publishers B.V., Amsterdam

Printed in The Netherlands

BOOK REVIEW

D. H. Sleeman and J. S. Brown (eds.), Intelligent Tutoring Systems. London: Academic Press, 1982, ISBN 0-12-648680-8, 345 pages, $36. This is an excellent collection of papers, some of which appeared in a special issue of the International Journal of Man-Machine Studies in 1979. Each addresses one or more fundamental problems in ITS, such as how to provide students with non-intrusive coaches and advisors (e.g., Burton and Brown, Goldstein), how to model a student's basic competence or incompetence in some domain (e.g., Sleeman, Burton), how to design friendly interfaces and supportive environments (e.g., Miller, Genesereth). Many well-known systems are described in some detail, e.g. WEST, WUMPUS, GUIDON, LMS, BUGGY and SOPHIE. Over the last ten years, ITS has grown out of generative CAI (Computer Aided Instruction) as researchers have attempted to provide more adaptive learning environments, systems which are more responsive to differences between students. Such systems include problem-solving monitors, coaches and consultants. They require, amongst other things, some model of the student, so that the misconceptions of individual learners can be diagnosed and corrected either on or off line. Sleeman and Brown point out in their introduction that ITS has abandoned one of CAI's early objectives, that of providing total courses. The emphasis now is upon building systems that deal with a fairly limited domain, such as place value arithmetic and simple algebra. Nevertheless, such topics are educational "watersheds," in the sense that these skills are crucial to many other topics met later in the curriculum, and so they merit special attention. This attention is provided by three papers in particular: Matz and Sleeman on high school algebra, and Burton on children's errors in subtraction. In each case, the authors are looking for a procedural model of student behaviour which allows systematic errors to be identified and remedied. Matz is concerned with the genesis of error through extrapolation, Sleeman reports experiments with the Leeds Modelling System, while Burton concentrates upon the decomposition of skills into subskills. Nevertheless, ITS is not confined to mathematical subject matter, as Stevens, Collins and Goldin show with their knowledge-based approach to geography. There is a close relationship between work on expert systems in Artificial Intelligence and work on intelligent tutoring systems. One of the things that distinguishes ITS from more conventional CAI is that the program that the student interacts with is generally capable of performing the task or solving the problem, rather than merely having access to a pre-stored solution. Clancey's paper provides some very interesting insights into the problems encountered when one attempts to build an ITS (GUIDON) around an extant expert system (MYCIN) which performs medical diagnoses. Two papers pay particular attention to the man-machine interface: Sleeman and

398 Hendley address some of the difficulties of providing natural language explanations in the ACE system, while Genesereth looks at ways of inferring a student's problem-solving plan for the purpose of giving help with an automatic advisor. WEST (Burton and Brown) and W U M P U S (Goldstein) are also concerned with how best to provide advice, while Miller's Spade-O system represents an attempt to build a structured environment in which beginners can acquire programming skills through a mixed-initiative dialogue. Brown, Burton and de Kleer's paper on S O P H I E I, II and III is full of practical insights into the techniques employed in implementing one of the best known examples of ITS. O'Shea and Kimble both describe self-improving tutors, i.e., Programs which are capable of improving their teaching performance as a result of interacting with students. For readers new to ITS, there is probably no better introduction to the field, while researchers in the area will find it useful to have these papers together in the same volume. Peter Jackson

Open University, Milton Keynes

BOOKS REVIEWED (VOLUME 12) Entwistle, Noel, Styles of Learning and Teaching: An Integrated Outline of Educational Psychologyfor Students, Teachers and Lecturers (Michael Eraut), 99 Jonassen, David H. (ed.), The Technology of Text: Principlesfor Structuring, Designing and Displaying Text (David Sless), 291 McMaster, J. McG., Methods in Social and Educational Caring (Will Swarm), 195 Rowntree, Derek, Developing Coursesfor Students (John D. Wilson), 191 Rowntree, Derek, Educational Technology in Curriculum Development (Richard N. Tucker), 192 Sleeman, D. H. and Brown, J. S. (eds.) Intelligent Tutoring Systems (Peter Jackson), 397 Stenhouse, Lawrence, et al., Teaching About Race Relations (Digby Anderson), 100