Making Learning Systems Work, edited by Diana Eastcott, Bob Farmer and Brian Lantz, pp226-239
Aspects of Educational and Training Technology Volume XXIII, Kogan Page, 1990
Moving information technology research from
the laboratory to the classroomChris F Reynolds, CODIL Language Systems Ltd, Tring,
Hertfordshire, UK
Abstract
Artificial intelligence and expert systems ideas are beginning to spill out of the laboratory but the first offerings tend to require powerful computers and sophisticated users. What is required is a reinterpretation of the basic ideas in a usable form for use in teaching environments. This paper discusses the steps needed to move new ideas from the computer laboratory to the classroom, using the MicroCODIL package as an example.
INTRODUCTION
Information technology (IT) is moving so fast that educational systems have difficulty in keeping up with it. Techniques which were current when the teacher was an undergraduate, are out of date after a few years in the classroom. Pupils at school today may well find that their knowledge is out of date when they start work. The matter is made worse by the natural resistance to change. For example, the desire for standardization means that Basic, a 25-year-old computer language from the first generation of high level languages, is still widely being taught to pupils who will be working into the 2040s.
Information technology (IT) is moving so fast that educational systems have difficulty in keeping up with it. Techniques which were current when the teacher was an undergraduate, are out of date after a few years in the classroom. Pupils at school today may well find that their knowledge is out of date when they start work. The matter is made worse by the natural resistance to change. For example, the desire for standardization means that Basic, a 25-year-old computer language from the first generation of high level languages, is still widely being taught to pupils who will be working into the 2040s.
There are a number of ways in which this problem can be tackled. Those linked with the management and funding of the educational system are beyond the scope of this paper. Instead it concentrates on the technical problems of transferring new developments from the research laboratories to the classroom, using a computer package called MicroCODIL as an example.
In trying to make such a transition there are a number of problems which need to be tackled.
The communication problem. Most research scientists use bad sophisticated technical language. Translating the research into everyday terms for the classroom is therefore a major task. What is available for transfer? Research scientists need to attract funds to do their work, and it would be fair to say that much of what is reported is of an optimistic nature. Something which supports carefully managed demonstrations in a laboratory may well be unable to support a group of inquisitive school children.Equipment costs. Most research is carried out with facilities which are far more generous than those available in the typical classroom. In some cases it may be impractical to transfer ideas until appropriate hardware becomes cheaper. In other cases it may be possible to develop tools which support an adequate teaching system by introducing techniques more relevant to low power computing.Trivializing the research. In reformatting research into a more amenable form, simplifications and assumptions will need to be made. Care must be taken not to simplify so far that the result is trivial, or not to simplify enough and end up with something which is too sophisticated for classroom use.
MICROCODIL
The MicroCODIL package consists of a language interpreter and a series of utilities which combine to provide a working environment for exploring and using a range of advanced information technology ideas in the classroom. It is supplied with over a dozen working knowledge bases, and a comprehensive manual (Reynolds 1987a).
The MicroCODIL package consists of a language interpreter and a series of utilities which combine to provide a working environment for exploring and using a range of advanced information technology ideas in the classroom. It is supplied with over a dozen working knowledge bases, and a comprehensive manual (Reynolds 1987a).
The MicroCODIL language is concerned with representing information in a flexible and easy-to-read manner, and with the minimum of restrictions. Its most notable feature is that there is no distinction between program and data. Everything is information. While the user can have files called MYDATA or MYPROGRAM, there are no arbitrary distinctions, and it is possible to search either file with equal ease. It provides considerable logical processing power, while avoiding the need for nested brackets that are such a headache in languages initially developed for research usage, such as LISP and PROLOG.
This flexible approach means that there is no need for predefined record structures (unless you want them). For instance a book may have one author, many authors or no author, and you do not need to know of these possibilities when asking a question. Ranges and approximate information can be stored and uncertain information can be qualified by probabilities. Properties can be inherited in a way that resembles object-oriented programming, so that 'AUTHOR-Smith'. will be found in a search for 'PERSON equal Smith'. In addition a 'fuzzy' interface allows the user to specify alternative definitions, ranging from a list of synonyms to complex logical deductions.This combination of facilities makes it possible to demonstrate a wide range of information technology concepts, including artificial intelligence and problem solving, and expert system style explorations, with a running commentary at each step. It can also handle sophisticated information storage and retrieval tasks (for details see Reynolds 1987a, 1988a).
The supplied knowledge bases cover a variety of subjects. BIRDS contains simple details of British birds at about primary school set theory level. The aim of BOGBOD is to discover the identity of the body in the bog. CRISIS is a fun item which provides an easy way to explore MicroCODIL's powerful diagnostic facilities. DEVON and PRISON use BASIC files to hold-historical data, which are interrogated using MicroCODIL. FRENCH contains a small dictionary, and a series on interactive lessons. It is designed to encourage pupils to think about the question what is information in a context most will easily understand. INVENT is another historical knowledge base, containing five inventories from between 1669 and 1709, using the original spelling! It demonstrates approximate matching, arithmetic in old money, and simple hierarchical definitions. KINGS is a retrieval package for younger children - but use of diagnostics will show that it uses artificial intelligence techniques to get some of the answers. MUSIC is included to show teachers, and older children, how to interface MicroCODIL to the BBC computer's hardware features, such as sound. PRICE illustrates how a non-procedural approach can give useful results in a business environment. Computer-aided learning techniques are demonstrated by QUEST, and new files of questions and answers may be added. Finally SOLIDS shows that even simple school chemistry involves sophisticated information processing tasks.
The biggest distributed knowledge base, FARMER allows searching of a file describing the farms and farmers of the parish of Sandridge, Herts, (see Reynolds 1987b, 1989a). It contains information from a wide variety of historical sources, and instructions are supplied to allow teachers to set up their own files of local history.
THE DESIGN OPTIONS
Deciding what can be done
It is not possible to take powerful research packages, designed for use by highly educated scientists, and simply transfer them to the classroom. A choice has to be made on what elements of the research need to be transferred. Various approaches are possible:
Teaching a subset of the programming skills used in current research. PROLOG subsets are sometimes used for this purpose. This approach may be suitable for specialist teaching but it is less suitable for general information technology appreciation.Developing a series of specialist 'black box' packages which demonstrate clever things. A chess playing program would fall into this category. This approach is useful to show pupils what computers can do, but there is a danger of the 'Gee Whiz' reaction where the less able are either frightened off or assume that computers can do anything.Produce a specific educational system which allows pupils to explore a range of relevant concepts for themselves while avoiding the problems of formal mathematical techniques. This is the approach followed in MicroCODIL (for non-procedural information processing) and LOGO (for handling geometrical and simple logical procedural tasks). This approach is the most suitable for allowing pupils to get a feel for what computers can do, while at the same time realizing that they can control what is going on
The last of these approaches is the one considered below, with special reference to MicroCODIL.
Simplifying the concepts
It is important that the basic concepts are as simple as possible, so that pupils have the minimum to learn and understand in order to get started. In the case of MicroCODIL a unified approach was taken towards 'information' rather than imposing the conventional approach of dividing tasks into two rigidly separate parts referred to as 'program' and 'data'. This means less to learn and fewer restrictions, as pupils do not have to worry about which is which. It also reflects research ideas in languages such as LISP, non-procedural systems, and in object-oriented programming. By making good use of the concept of recursion it has been possible to avoid the problem of nested brackets that cause real headaches with languages such as PROLOG and LISP.
Some of the techniques used are novel and details are being published elsewhere (see Reynolds, 1989b).
It is important that the basic concepts are as simple as possible, so that pupils have the minimum to learn and understand in order to get started. In the case of MicroCODIL a unified approach was taken towards 'information' rather than imposing the conventional approach of dividing tasks into two rigidly separate parts referred to as 'program' and 'data'. This means less to learn and fewer restrictions, as pupils do not have to worry about which is which. It also reflects research ideas in languages such as LISP, non-procedural systems, and in object-oriented programming. By making good use of the concept of recursion it has been possible to avoid the problem of nested brackets that cause real headaches with languages such as PROLOG and LISP.
Some of the techniques used are novel and details are being published elsewhere (see Reynolds, 1989b).
User friendly approach
A typical research system is designed to demonstrate sophisticated ideas to academics. When used as a tool to support a university-based research or development task there will be high motivation on the part of the users to get
results. As a result many research systems have very poor user interfaces. Such a limited approach is totally unacceptable in the classroom. What is needed is an attractive interface, and reasonably fast result. The general aspect of the system should help to encourage and motivate pupils, who will compare it with the highly sophisticated user interfaces of modern computer games.
A typical research system is designed to demonstrate sophisticated ideas to academics. When used as a tool to support a university-based research or development task there will be high motivation on the part of the users to get
results. As a result many research systems have very poor user interfaces. Such a limited approach is totally unacceptable in the classroom. What is needed is an attractive interface, and reasonably fast result. The general aspect of the system should help to encourage and motivate pupils, who will compare it with the highly sophisticated user interfaces of modern computer games.
In MicroCODIL, a wide range of human computer interaction features were included in the design (see Reynolds 1987c). These included using colour (see Reynolds 1988b), ensuring that the display was as flicker free as possible, and ensuring that when a single operation takes more than a couple of seconds (for example searching for information stored on a large file on disc) a warning message is displayed.
Good working environment
In moving from a research laboratory to a school classroom there is a far greater need for an integrated working environment. In the case of MicroCODIL, this includes the provision of an editor, a sort package and other tools. In
addition comprehensive help and diagnostic facilities are needed to help the pupil to learn the system, and understand their errors. The provision of such support tools actually represents a significant part of the overall package.
In moving from a research laboratory to a school classroom there is a far greater need for an integrated working environment. In the case of MicroCODIL, this includes the provision of an editor, a sort package and other tools. In
addition comprehensive help and diagnostic facilities are needed to help the pupil to learn the system, and understand their errors. The provision of such support tools actually represents a significant part of the overall package.
Finding suitable examples
One of the big problems in introducing a completely new language system such as MicroCODIL is in finding suitable examples. There is the need to teach the teacher about the system, and how it can be used in the classroom across the curriculum. A compromise is needed to include a wide range of language features and teaching environments, while at the same time keeping the total number of examples within manageable proportions.
Support documentation
Because of the wide range of potential users, documentation is perhaps the biggest headache. At one end of the scale there is the computer science teacher writing new commands to drive a speech synthesizer. In the middle there is the history teacher who wants to set up a local history knowledge base but knows very little about practical computing. At the other end there is the pupil running one of the exercises provided with the package as part of his or her classroom teaching.
Because of the wide range of potential users, documentation is perhaps the biggest headache. At one end of the scale there is the computer science teacher writing new commands to drive a speech synthesizer. In the middle there is the history teacher who wants to set up a local history knowledge base but knows very little about practical computing. At the other end there is the pupil running one of the exercises provided with the package as part of his or her classroom teaching.
Getting new technology across can also be difficult. Some concepts are quite easy to explain to the virgin mind, but could well clash with concepts that the pupil (or sometimes even the teacher) has assimilated from the use of BASIC, etc. A history teacher wants to use inexact and incomplete data because that is the raw material of history, and is not interested in the theoretical background which is important to the computer science teacher. Ideally there would be different manuals and instruction books for different users, but this kind of luxury is not economically possible until a package is widely accepted.
Fitting into the hardware
The MicroCODIL software had run on a minimal 32K BBC computer with a single 40-track floppy disc. This acted as a major spur to get the design right (on a big machine it is often possible to write vast quantities of program code to get around problems). Unfortunately it also meant a number of major compromises had to be made. For example, the edit and sort facilities are separate programs, accessible from menus, rather than a fully integrated part of the MicroCODIL interpreter. In a different area, it was decided that it was more important to be able to demonstrate approximate matching and probabilities than to provide a sophisticated report output, as the latter is widely available on conventional software packages.
The MicroCODIL software had run on a minimal 32K BBC computer with a single 40-track floppy disc. This acted as a major spur to get the design right (on a big machine it is often possible to write vast quantities of program code to get around problems). Unfortunately it also meant a number of major compromises had to be made. For example, the edit and sort facilities are separate programs, accessible from menus, rather than a fully integrated part of the MicroCODIL interpreter. In a different area, it was decided that it was more important to be able to demonstrate approximate matching and probabilities than to provide a sophisticated report output, as the latter is widely available on conventional software packages.
CONCLUSIONS
Transferring current research from the laboratory into the classroom is not an easy task, and compromises are necessary in order to simplify the concepts and fit them into a teaching environment. This is particularly true where the aim is to provide a comprehensive knowledge-based system such as MicroCODIL. While the results are extremely useful, there is much that could be done to produce a better system, especially if the approach is applied to the more powerful systems that are now becoming available in schools. In particular the tasks of designing a user-friendly support environment, with comprehensive tested examples and documentation, represent a very significant part of the development costs. One has only to look at the number of secondary schools in the United Kingdom, and ask how much they are prepared to pay for a software package (assuming that every school bought a copy) to realize that such developments will not occur if they are required to be economically viable from an early stage.
Transferring current research from the laboratory into the classroom is not an easy task, and compromises are necessary in order to simplify the concepts and fit them into a teaching environment. This is particularly true where the aim is to provide a comprehensive knowledge-based system such as MicroCODIL. While the results are extremely useful, there is much that could be done to produce a better system, especially if the approach is applied to the more powerful systems that are now becoming available in schools. In particular the tasks of designing a user-friendly support environment, with comprehensive tested examples and documentation, represent a very significant part of the development costs. One has only to look at the number of secondary schools in the United Kingdom, and ask how much they are prepared to pay for a software package (assuming that every school bought a copy) to realize that such developments will not occur if they are required to be economically viable from an early stage.
REFERENCES
Reynolds, C. F. (1987a) MicroCODIL Manual (and associated software). CODIL Language Systems, Tring.
Reynolds, C. F. (1987b) MicroCODIL and History. CODIL Language Systems, Tring
Reynolds, C. F. (1987c) Human factors in systems design. In Diaper, D. and Winder, R. (eds) People and Computers, III. Cambridge University Press.
Reynolds, C. F. (1988a) Introducing expert systems to pupils. Journal of Computer Assisted Learning, 4, 79-92.
Reynolds, C. F. (1988b) The use of colour in language syntax analysis. Software Practice and Experience, 17,513-519.
Reynolds, C. F. (1989a) A flexible approach to local history data bases. In Proceedings of the International Conference on Computer in the History Classroom, University of Leeds.
Reynolds, C. F. (1989b) CODIL, the architecture of an information language. Computer Journal, (in press).
Reynolds, C. F. (1987b) MicroCODIL and History. CODIL Language Systems, Tring
Reynolds, C. F. (1987c) Human factors in systems design. In Diaper, D. and Winder, R. (eds) People and Computers, III. Cambridge University Press.
Reynolds, C. F. (1988a) Introducing expert systems to pupils. Journal of Computer Assisted Learning, 4, 79-92.
Reynolds, C. F. (1988b) The use of colour in language syntax analysis. Software Practice and Experience, 17,513-519.
Reynolds, C. F. (1989a) A flexible approach to local history data bases. In Proceedings of the International Conference on Computer in the History Classroom, University of Leeds.
Reynolds, C. F. (1989b) CODIL, the architecture of an information language. Computer Journal, (in press).
No comments:
Post a Comment