Over the last four years, an increasing number of final year degree students in Leeds Polytechnic's Division of Information Technology have chosen for their required project to work within the general theme of instructional technology for student centred learning.
This is particularly appropriate at a time when Leeds and other Polytechnics are attempting a massive expansion in their student numbers and, since this increase is not being matched by corresponding increases in staffing, are looking towards technological solutions to help offer students the ability to take responsibility for and to some extent conduct their own learning
This paper will summarise the aims, methodologies, results and conclusions of twelve such projects completed during the period 1988-1991. These projects include development of working prototype systems involving hypermedia, intelligent tutoring systems, educational knowledge based systems, and interactive video.
Furthermore, many of these projects have involved the interlinking of such systems within a multimedia context, and the experiences of the researchers in attempting to achieve this integration of different presentation modes within a single teaching package will be discussed.
Furthermore, the text based materials may link through a hypertext system with any of the other media and cause them to be accessed automatically as the student explores the domain in a student centred, goal oriented fashion.
Our Faculty at Leeds has for some five years been particularly interested in the possibilities posed by Interactive Video and its interlinking with other packages, such as database, expert system shell, authoring language and hypertext. The ultimate aim might be to allow all of these to inter-communicate via the controlling mechanism of an educational Adviser.
A large part of the work has so far been conducted during the last three years through a dozen or so final year projects undertaken by undergraduates on the Faculty's BSc Computing who have chosen to work in this area. That students have selected this area may be attributed in large measure to an increased level of interest in Artificial Intelligence and Instructional Technology fostered within the Faculty by the authors who were directly involved in acting as supervisors or advisers for many of these projects.
TOPCLASS and CRYSTAL were each used to create a teaching program to the same specification using the BBC's 'Learn Golf' videodisc, and working demonstrator programs were produced for comparison. Dobby found that CRYSTAL was easy to learn and program, that it offered structured program design, and that It could supply explanations. However, his criticisms were that the communication link was in one direction only (from computer to interactive video player), that CRYSTAL had limited user response handling/checking, and that it offered no graphics capability.
The advantages of TOPCLASS were a two way communication link between computer and player, advanced user response handling, the availability of graphics and text overlays and animated graphics. Nevertheless, Dobby found it was not as easy to learn and program as was CRYSTAL. In conclusion he felt that neither system was Ideally suited to controlling Interactive Video.
Thomason (1989) constructed a student model to be used in conjunction with a knowledge base linked to an Interactive Video system. Having reviewed available video discs, 'Play Golf' was again chosen and mapped onto a tree structure domain representation. A designer toolkit was then written in Pascal for creation and manipulation of the domain, and a tutorial comprising questions and feedback was then produced using an 'in order' traversal of the tree which referenced a pre-requisite list for each node in order finally to access and play an appropriate segment of video from videodisc.
The role of student modelling in an intelligent Interactive Videodisc tutoring system was taken further by Perkin (1990), who aimed to create an 'intelligent' IV tutoring system by incorporating a comprehensive student model representing the current state of the student's knowledge as well as their preferred learning styles and previous relevant experience.
Suitable representations for the knowledge base and student model to be used in conjunction with a chromatography videodisc were designed as Pascal data files. Search algorithms were also written in Pascal and a system 'shell' produced. The resulting system allows the user manually to select a topic and receive a list of advised materials (video, audio or text). The video clips may then be viewed after which a test is administered and the percentage of correct responses Is presented to the user.
Hussein (1990) made a study into the use of Interactive Video as a learning aid in comparison with traditional methods. He wished to ascertain whether IV could be superior to traditional methods of learning. A linear video disc (Play Golf) was taken as a basis for writing a Pascal program to allow the disc to be used interactively. It was also used to produce a textual exposition of the same material.
A study was carried out to analyse students into types according to their choice of learning sequence. 'Efficient' (in terms of access timings) tutorial layouts were constructed and a Pascal program was written to simulate the search times for each student's most efficient sequence compared with that of the current lecturer.
Considerable variations in desired sequencing of topics were noted, but they fell Into a limited number of possible groupings. Discriminant analysis identified seven groupings within the seventeen subjects. Subjects working with a tape sequenced for their grouping would typically experience (according to the simulation) considerable and significant reductions in search times.
Chow (1990) also investigated methods to improve access time for an Interactive Video tape data retrieval system. His aim was to determine whether 'intermediate searching' (conducted while data are being entered) provides quicker data retrieval than 'final searching' begun after all search criteria have been entered.
Video footage of used cars (potentially for sale) was shot and edited into a hierarchically structured sequence (make, model, etc). A dBase IV program was written to input the user's requirements (constraints) and initiate searching to the start of the relevant section (make, model, etc). Two comparisons were included in the studies. A structured version was created which initiated searching only after all constraints had been entered, and an unstructured randomly sequenced tape of the cars.
User studies were employed to measure retrieval times under intermediate and final search regimes, and to determine user attitudes. As a result, search time was not in f act found to be improved by the intermediate search technique. Nor was it found to be improved for a single direct search by restructuring the material on the tape.
Kay points out that the browsing facility which is central to the concept of hypertext is potentially less advantageous within a tutoring environment where there Is an element of conflict between allowing the student to follow a personally meaningful pattern of knowledge acquisition whilst at the same time preventing him from following a tangled and ultimately meaningless route through the domain. He represented the complex and variable relationships between the techniques within his domain in a meaningful way via the, sequencing mechanism referred to as a 'dynamic web'. The dynamic web provides a mechanism through which the student can be prevented from straying from the principal area of study by restricting the links and therefore the movement of the student within the domain.
Fendley's (1988) approach in building a knowledge based Advisory system was to explore the use of a planner and knowledge domain to aid self generation of teaching routes. He also incorporated a student modeller to cater for individual student parameters.
He chose to represent the knowledge domain through a 'Directed Acyclic Graph' and then filled it out with an example subject domain (Descriptive Statistics). A student model was then set up using Goldstein's (1982) method of overlaying the knowledge domain. Finally a 'planner' was developed (written in Pascal) based on the specification given by Peachy and McCalla (1986). The result was a working demonstrator program capable of selecting an appropriate sequence of topics for the learner together with individualised teaching advice on each topic.
For each topic node in the domain of Fendley's system, the planner considers whether the node's post-conditions contribute towards the overall learning goal as it currently exists, and whether the node's prerequisite nodes are at a sufficient level of understanding for it to be 'eligible' for selection. The planner also makes use of backtracking techniques in an attempt to find an optimal route through the subject domain. Additional code was written for the next stages of 'route evaluator' and 'plan executor'.
Wong (1990) wished to investigate possible search strategies and mechanisms for use in the Adviser. She chose a tree representation for the knowledge domain, using an example domain of 'Pascal types'. Like Fendley, she used an overlay student model to record current topic strengths, preferred teaching style, learner type, time available and confidence. Possible candidate rules were formulated for navigating the domain for the purpose of deriving a sequence of learning advice.
A working system for generating advice route traces was produced, and the topic sequence traces It produced were evaluated by a lecturer who largely approved them. A set of design tools was constructed in Pascal for creating and modifying the domain and student model, and some of the proposed rules were coded into a navigation program. Student stereotype data was fed into the navigation program which could then produce traces of the corresponding advice sequences.
In contrast, Morris (1990) concentrated on the interface between knowledge engineer and Adviser by investigating methods for improving their interaction using graphical tools. After familiarisation with the Adviser Hobbs (1990) was developing and with the graphics facilities offered by the language Smalltalk (in which Hobbs was constructing it), he wrote additional code in order to provide a graphical interface.
For example, one modification he introduced concerned the existing conventional menu interaction for changing the domain structure (eg by adding/deleting nodes). This was replaced by an interaction employing windows, popup menus and 'click on' selection, thereby reducing keystrokes and allowing alternative views on the domain simultaneously.
He reviewed the existing representational style of the knowledge domain In comparison with other styles he proposed, one of which he set up and used in evaluation studies with users who also helped appraise the two alternative interaction modes. He found firstly that the windows interface was generally liked better than the previous Interface, and secondly that his alternative representation was felt to be better at representing secondary links but otherwise less easy to work with than the tree representation.
Dobby, D. (1989). A comparison of methods for controlling an Interactive Video [Unpublished project report].
Fendley, C. (1988). An Intelligent knowledge based educational Adviser [Unpublished project report]
Goldstein, I. P. (1982). The genetic graph: A representation for the evolution of procedural knowledge. In Sleeman, D. and Brown, J. S. (eds), Intelligent Tutoring Systems. Harcourt Brace Jovanovich.
Hobbs, D. J. (1988). Design of a knowledge based educational Adviser. Programmed Learning and Educational Technology, 25(4).
Hobbs, D. J. (1990). Second level design of a knowledge based educational Adviser. Educational and Training Technology International, 27(2).
Hussein, A. (1990). A study into the use of Interactive Video as a learning aid and comparison with traditional methods [Unpublished project report].
Morris, K. (1990). Design and evaluation of methods for interacting graphically with the knowledge domain in a knowledge based educational Adviser [Unpublished project report].
Peachy, D. R. and McCalla, G. I. (1986). Using planning techniques in intelligent tutoring systems. International journal of Man-Machine Studies, Vol 24.
Perkin, R. (1990). The role of student modelling in an intelligent Interactive Videodisc tutoring system [Unpublished project report].
Platts, I. (1989). A comprehensive study of Interactive Video technology with particular reference to the design of an efficient IV tape tutorial system [Unpublished project report].
Thomason, I. (1989). Student modelling in a computer assisted Interactive Video environment [Unpublished project report].
Wong A (1990), An investigation of possible search strategies and mechanisms for use in a knowledge based educational Adviser [Unpublished project report].
|Authors: Dave Hobbs and David Moore are Senior Lecturers in the Faculty of Information and Engineering Systems, Leeds Polytechnic, UK.
Please cite as: Hobbs, D. and Moore, D. (1992). Instructional technology for student centred learning: The Leeds Polytechnic experience. In Promaco Conventions (Ed.), Proceedings of the International Interactive Multimedia Symposium, 227-234. Perth, Western Australia, 27-31 January. Promaco Conventions. http://www.aset.org.au/confs/iims/1992/hobbs.html