- Email: [email protected]
Cognitively adapted hypertext for learning
Symbiosis of Human and Artifact Y. Anzai, K. Ogawa and H. Mori (Editors) © 1995 Elsevier Science B.V. All rights reserved.
COGNITIVELY
ADAPTED
HYPERTEXT
301
FOR LEARNING
Kelvin Clibbon LUTCHI Research Centre, D e p a r t m e n t of Computer Studies Loughborough University of Technology Loughborough, LE 11 3TU UK Email [email protected] Telephone +44 509 222327
ABSTRACT This paper discusses the effect of adapting hypertext to the learner. Cognitive overhead and disorientation limit the effectiveness of hypertext for learning. By cognitively a d a p t i n g a h y p e r t e x t s y s t e m to the u s e r and by providing instructional cues, the effects of these problems might be reduced. A quasiexperimental evaluation study is reported, with a view to testing the efficacy of this theory. 1. INTRODUCTION The t e r m h y p e r t e x t eludes a simple definition, but for the purpose of this research, it is the technology of non-sequential r e a d i n g and writing [1]. A hypertext consists of linked nodes. A node is a discrete unit of text, graphics, sound, or whatever; within which there are links to other nodes. A link is a connection between parts of text or other material. Links provide the structure of a h y p e r t e x t system and act as pointers from one node to another. This linking capability allows text to be organised in a non-linear fashion. The basic premise behind educational hypertext is t h a t learners are free to forge their own paths through the richly interconnected information space in a self-directed manner. Thus satisfying their own educational goals, r a t h e r t h a n h a v i n g to slavishly follow some form of l i n e a r t u t o r i a l [2]. The m a i n characteristics of hypertext systems t h a t have potential for learning are: they support the collection and structuring of information; they are enabling, in the sense t h a t they allow high levels of learner control; and they have the potential to alter the roles of teachers and learners, and the critical interaction between them.
2. PROBLEMS WITH HYPERTEXT FOR LEARNING Hypertext however, is a two-edged sword. On the one h a n d it provides access to flexibly structured information, but on the other, its very flexibility can lead to problems. The major problems are the difficulties a s s o c i a t e d with the a u t h o r i n g of h y p e r t e x t systems, and the difficulty of navigating through the information space.
302 The "authoring problem" [3] occurs because hypertext links reflect what the author thinks are important, more often than they reflect what the reader hopes to find. The authoring process depends on the way the author understands the structure and flow of the material. There are as of yet no established guidelines for authoring educational hypertext documents. "Lost in Hyperspace" [4] refers to the feeling of disorientation a reader can experience when following a connected trail of hypertext links. The major cause of disorientation is cognitive overload. When many nodes and links must be examined, user's may not know how to reach a specific node in the network or what to examine next, or may have even forgotten what information they were seeking originally. This leads to what has been called "cognitive overhead ..... the additional effort and concentration necessary to maintain several tasks or trails at one time" [4]. The disorientation problem has implications for learning from hypertext. Halasz [5] states that completely free exploration of a network of nodes and links will be sub-optimal for learning. Learners will fail to gain an overview of the material and have difficulty in locating information they know is available. The authoring and navigation problems are not mutually exclusive. If the author's view does not concur with the information needs of the reader, then the author created links will be of little value. The author's network might be insufficiently rich for the user's needs; meaning that they will have to add links to make the document usable - if node and link editing facilities are available. Although this may be beneficial in educational hypertext systems, editing the hyperdocument would increase cognitive overhead and is at best inconvenient, and at worst impractical. Advocates of hypertext-based learning systems are introducing tools for navigation and guidance (e.g. [6] ). Such extensions are forms of soft tutoring, as opposed to the strict model-driven tutoring that characterises Intelligent Tutoring Systems (e.g. [7]). A contrasting way of helping learners to choose, is to give them the option of relinquishing control for some activities. The use of tours is one such option. Another option is used with StrathTutor [8]. Learners are required to make sense of dynamic hypertext links computed by the system, on the basis of knowledge attributes of displayable information. The types of control in hypertext learning systems consist of control over the sequencing of materials, or control of the learning activities, such as browsing, trying interactive demonstrations or problem solving. The optimal level of control depends on the nature of the learner, on their familiarity with the materials, on their learning goals and on the nature of the knowledge domain [9]. A key research issue is in determining the appropriate level of instructional control. 3. C O N C E P T U A L L Y A D A P T E D H Y P E R T E X T F O R L E A R N I N G
This paper reports on a project whose aim was to develop a methodology for authoring adaptive educational hypertext systems. Such systems should provide support for learning activities and have adaptive links. Link adaptability was supported in a prototype system by an underlying conceptual
303 model of the domain (structured programming techniques for this case study). Concept tutoring techniques, based on instructional design theory, were used to develop the conceptual model (see Merrill et al., [10]). Hypertext links are only provided to the user if the required prior nodes, containing links to worked examples and quiz's for the student, have been visited. The students are also provided with instructional cues - for guidance at strategic points of the interaction, to ensure t h a t they are given the opportunity to explore the important nodes in the hyperspace. The aim is to reduce the cognitive overhead experienced by students at the outset of the session, increasing the n u m b e r of links available, as they become more familiar with the domain. Each of the hypertext nodes has an attached set of production rules [11], which infer what links to provide to the learner, depending on their interactions and the conceptual model. 4. E V A L U A T I N G A D A P T E D H Y P E R T E X T F O R L E A R N I N G
To test the efficacy of this approach, a quasi-experimental evaluation study was carried out. The major research issue was to investigate w h e t h e r conceptually adapting the hyperdocument and providing instructional cues enhances learning outcome. Two hyperdocuments containing the subject m a t e r i a l to be learned were developed. One document was the already described conceptually a da pt e d hypertext. The other d o c u m e n t was a conventional static hypertext, providing the learner with a parsimonious set of links to facilitate exploration. A conventional lecture was used as a control presentation method. A total of 85 students participated in the study. Students completed an elementary structured programming test (to ascertain their prior knowledge level of the domain to be taught), and were categorised into one of Honey and Mumford's [12] four learning styles - Activists, Reflectors, Theorists and Pragmatists. It is believed that individual differences interact with learning from hypertext. Not all learners are equally apt in learning from hypertext environments; student's generally have different learning styles. A learning style is part of an individual's cognitive structure and refers to the pattern of behaviour and performance by which an individual approaches educational or learning experiences. The term learning style is used as a description of the a t t r i b u t e s and behaviours which determine an individuals preferred way of learning [12]. Individuals approach the learning of new ideas and concepts differently. Esichaikul et al. [13] advocate that authors need to take into account learning styles when designing hyperdocuments. For example, systems could include paths t h a t complement or accommodate the preferred learning style of a particular learner. The students were sequentially assigned to one of three t r e a t m e n t groups, each using a different t r e a t m e n t method (a lecture, a cognitively adapted hypertext or a static hypertext). Those students allocated to the lecture group were given a presentation and guided through some worked examples, lasting a total of one hour. Students in the cognitively adapted hypertext and the static
304 hypertext groups spent one hour learning as much of the information in the s y s t e m s as possible. Upon finishing t h e i r respective sessions, s t u d e n t s completed a test to ascertain the a m o u n t of material learned. The learning outcome was m e a s u r e d by a fourteen question post-test, which resulted in a single score. Questions were developed to test u n d e r s t a n d i n g of the key concepts and basic procedures. 5. R E S U L T S
Three research questions where addressed: (1) Does L e a r n i n g Style affect l e a r n i n g outcome? (2) Does the P r e s e n t a t i o n Method (Lecture, Cognitively Adapted Hypertext or Static Hypertext) affect learning outcome? (3) Are there interaction effects between learning styles and method of p r e s e n t a t i o n t h a t affect learning outcome? The dependent variable used in the study was an objective m e a s u r e m e n t of learning outcome. The independent variables include: (a) the learning style of the l e a r n e r and (b) the method of p r e s e n t a t i o n of the material. U n i v a r i a t e analyses of variance were used to test for significant differences in the means of the learning outcome scores between the different methods and the learning styles. Specifically, a two-factor ANOVA was used to investigate interactions effects between presentation method and learning styles. The learning outcome score means for each of the study design groups are presented in table 1, along with the number of subjects in each group and the standard deviation. Table 1. Learning Outcome by Learning Style and Presentation Method
Activists; means s.d. n
Reflectors; means s.d. n
Theorists; means s.d. n
Pragmatists; means s.d. n
Method Totals; means s.d. n
Lecture
Adaptive H~,pertext
Static H~cpertext
Learning Style Totals
18.90 6.74 10
17.67 3.85 12
14.50 4.03 10
17.06 5.15 32
19.11 2.57 9
16.27 3.20 11
14.10 5.39 11
16.32 4.36 31
19.00 1.73 3
17.75 1.71 4
12.00 3.92
16.00 4.05 11
17.50 4.51 4
21.00 6.24 3
16.50 3.42 4
18.10 4.59 11
18.77 4.63 26
17.50 3.74 30
14.28 4.47 29
16.79 4.63 85
4
,,
305 The differences among the means for the learning outcome scores across the three presentation methods were significant. (F2,82 = 8.2, p = 0.0006). Multiple range tests, indicated t h a t the static hypertext session m e a n is significantly different from the lecture and the cognitively adapted hypertext session at the 0.05 level. This separation of means is highlighted in figure 1. The differences among the mean scores between learning styles (F3,81 = 0.529, p = 0.664) were not significantly different and there was no significant i n t e r a c t i o n effect between learning style and presentation method on learning outcome.
21 4
20 19 18 17 6 -
5
=
[ 14. Lecture
13
Adaptive hypertext 12
D - Static h y p e r t e x t
Activists
Reflectors
Theorists
Pragmatists
F i g u r e 1. L e a r n i n g O u t c o m e P r o f i l e s for P r e s e n t a t i o n M e t h o d 6. C O N C L U S I O N This research a t t e m p t s to reduce the cognitive overhead and disorientation experienced by l e a r n e r s while p u r s i n g t h e i r i n s t r u c t i o n a l objectives in browsing a hyperdocument. Instructional cueing makes the conceptual model incorporated in the adaptive system explicit to the learner. This study reports the effects a conceptually adapted hypertext and an individuals learning style, have on learning outcomes with a hypertext system. The mean scores of those subjects who attended the lecture (mean = 18.8) and used the adaptive hypertext
306 (mean = 17.5) were significantly different from those who used the static hypertext (mean = 14.3). This would seem to indicate that students learn from an adaptive hypertext incorporating instructional cues, to the same level as students attending lectures, for the domain in question. Future research will determine how the methodology described herein generalises to other domains. The main result from the study is that there was no significant difference in learning outcome between the cognitively adapted hypertext system and the conventional lecture. There was however a significant difference between these two presentation methods and the static hypertext system. Hence adapting the hypertext document and providing instructional cues to the learner may reduce cognitive overhead and increase learning outcome. REFERENCES
1. 2. 3. 4. 5. 6.
7. 8. 9. 10. 11. 12. 13.
J. Nielsen, Hypertext and Hypermedia, San Diego, CA: Academic Press, 1990. W. Reader and N. Hammond, Computer-Based Tools to Support Learning from Hypertext: Concept Mapping Tools and Beyond, Computers and Education, Vol. 22, Part 1/2, pp 99-106 (1994). C. Tompsett, Knowledge-Based Support for Hypertext, 5th International Expert Systems Conference, Learned Information: Woodside: pp 31-43, (1989). J. Conklin, Hypertext: An Introduction and Survey, IEEE Computer, Vol. 20, Part 9, pp 17-41. (1987). F. Halasz, Reflections on NoteCards: Seven Issues for the Next Generation of Hypermedia Systems, Communications of the ACM, Vol. 31, Part 7, pp 836-852 (1988). T. Mayes, M. Kibby and T. Anderson, Signposts for Conceptual Orientation: Some Requirements for Learning from Hypertext, In R. McAleese and C. Green (Eds.), Hypertext: State of the Art, Intellect, pp 121129, 1990. T. Sokolnicki, Towards Knowledge-Based Tutors: A Survey and Appraisal of Intelligent Tutoring Systems, Knowledge Engineering Review, Vol. 6, Part 2, pp 54-95 (1991). T. Mayes, M. Kibby and H. Watson, StrathTutor: The Development and Evaluation of a Learning-by-Browsing System on the Macintosh, Computers and Education, Vol. 12, Part 1, pp 221-229 (1988). N. Hammond, Teaching with Hypermedia: Problems and Prospects. In H. Brown (Ed.) Hypermedia/Hypertext and Object-Oriented Databases. Chapman and Hall, pp 107-124, 1991. M. Merrill, R. Tennyson and L. Posey, Teaching Concepts: An Instructional Design Guide (2nd Edition), Educational Technology Publications, Englewood Cliffs, New Jersey, 1992. T.Williams and B. Bainbridge, Rule Based Systems, In G. Ringland and D. Duce (Eds.), Approaches to Knowledge Representation: An Introduction, John Wiley and Sons, pp 101-115, 1988. P. Honeyand P. Mumford, The Manual of Learning Styles, Honey UK, 1992. V. Esichaikul, R. Smith and G. Madey, The Impact of Learning Styles o n Problem-Solving Performance in a Hypertext Environment, Hypermedia Vol. 6, Part 2, pp 101-110 ( 1994).
COGNITIVELY
ADAPTED
HYPERTEXT
301
FOR LEARNING
Kelvin Clibbon LUTCHI Research Centre, D e p a r t m e n t of Computer Studies Loughborough University of Technology Loughborough, LE 11 3TU UK Email [email protected] Telephone +44 509 222327
ABSTRACT This paper discusses the effect of adapting hypertext to the learner. Cognitive overhead and disorientation limit the effectiveness of hypertext for learning. By cognitively a d a p t i n g a h y p e r t e x t s y s t e m to the u s e r and by providing instructional cues, the effects of these problems might be reduced. A quasiexperimental evaluation study is reported, with a view to testing the efficacy of this theory. 1. INTRODUCTION The t e r m h y p e r t e x t eludes a simple definition, but for the purpose of this research, it is the technology of non-sequential r e a d i n g and writing [1]. A hypertext consists of linked nodes. A node is a discrete unit of text, graphics, sound, or whatever; within which there are links to other nodes. A link is a connection between parts of text or other material. Links provide the structure of a h y p e r t e x t system and act as pointers from one node to another. This linking capability allows text to be organised in a non-linear fashion. The basic premise behind educational hypertext is t h a t learners are free to forge their own paths through the richly interconnected information space in a self-directed manner. Thus satisfying their own educational goals, r a t h e r t h a n h a v i n g to slavishly follow some form of l i n e a r t u t o r i a l [2]. The m a i n characteristics of hypertext systems t h a t have potential for learning are: they support the collection and structuring of information; they are enabling, in the sense t h a t they allow high levels of learner control; and they have the potential to alter the roles of teachers and learners, and the critical interaction between them.
2. PROBLEMS WITH HYPERTEXT FOR LEARNING Hypertext however, is a two-edged sword. On the one h a n d it provides access to flexibly structured information, but on the other, its very flexibility can lead to problems. The major problems are the difficulties a s s o c i a t e d with the a u t h o r i n g of h y p e r t e x t systems, and the difficulty of navigating through the information space.
302 The "authoring problem" [3] occurs because hypertext links reflect what the author thinks are important, more often than they reflect what the reader hopes to find. The authoring process depends on the way the author understands the structure and flow of the material. There are as of yet no established guidelines for authoring educational hypertext documents. "Lost in Hyperspace" [4] refers to the feeling of disorientation a reader can experience when following a connected trail of hypertext links. The major cause of disorientation is cognitive overload. When many nodes and links must be examined, user's may not know how to reach a specific node in the network or what to examine next, or may have even forgotten what information they were seeking originally. This leads to what has been called "cognitive overhead ..... the additional effort and concentration necessary to maintain several tasks or trails at one time" [4]. The disorientation problem has implications for learning from hypertext. Halasz [5] states that completely free exploration of a network of nodes and links will be sub-optimal for learning. Learners will fail to gain an overview of the material and have difficulty in locating information they know is available. The authoring and navigation problems are not mutually exclusive. If the author's view does not concur with the information needs of the reader, then the author created links will be of little value. The author's network might be insufficiently rich for the user's needs; meaning that they will have to add links to make the document usable - if node and link editing facilities are available. Although this may be beneficial in educational hypertext systems, editing the hyperdocument would increase cognitive overhead and is at best inconvenient, and at worst impractical. Advocates of hypertext-based learning systems are introducing tools for navigation and guidance (e.g. [6] ). Such extensions are forms of soft tutoring, as opposed to the strict model-driven tutoring that characterises Intelligent Tutoring Systems (e.g. [7]). A contrasting way of helping learners to choose, is to give them the option of relinquishing control for some activities. The use of tours is one such option. Another option is used with StrathTutor [8]. Learners are required to make sense of dynamic hypertext links computed by the system, on the basis of knowledge attributes of displayable information. The types of control in hypertext learning systems consist of control over the sequencing of materials, or control of the learning activities, such as browsing, trying interactive demonstrations or problem solving. The optimal level of control depends on the nature of the learner, on their familiarity with the materials, on their learning goals and on the nature of the knowledge domain [9]. A key research issue is in determining the appropriate level of instructional control. 3. C O N C E P T U A L L Y A D A P T E D H Y P E R T E X T F O R L E A R N I N G
This paper reports on a project whose aim was to develop a methodology for authoring adaptive educational hypertext systems. Such systems should provide support for learning activities and have adaptive links. Link adaptability was supported in a prototype system by an underlying conceptual
303 model of the domain (structured programming techniques for this case study). Concept tutoring techniques, based on instructional design theory, were used to develop the conceptual model (see Merrill et al., [10]). Hypertext links are only provided to the user if the required prior nodes, containing links to worked examples and quiz's for the student, have been visited. The students are also provided with instructional cues - for guidance at strategic points of the interaction, to ensure t h a t they are given the opportunity to explore the important nodes in the hyperspace. The aim is to reduce the cognitive overhead experienced by students at the outset of the session, increasing the n u m b e r of links available, as they become more familiar with the domain. Each of the hypertext nodes has an attached set of production rules [11], which infer what links to provide to the learner, depending on their interactions and the conceptual model. 4. E V A L U A T I N G A D A P T E D H Y P E R T E X T F O R L E A R N I N G
To test the efficacy of this approach, a quasi-experimental evaluation study was carried out. The major research issue was to investigate w h e t h e r conceptually adapting the hyperdocument and providing instructional cues enhances learning outcome. Two hyperdocuments containing the subject m a t e r i a l to be learned were developed. One document was the already described conceptually a da pt e d hypertext. The other d o c u m e n t was a conventional static hypertext, providing the learner with a parsimonious set of links to facilitate exploration. A conventional lecture was used as a control presentation method. A total of 85 students participated in the study. Students completed an elementary structured programming test (to ascertain their prior knowledge level of the domain to be taught), and were categorised into one of Honey and Mumford's [12] four learning styles - Activists, Reflectors, Theorists and Pragmatists. It is believed that individual differences interact with learning from hypertext. Not all learners are equally apt in learning from hypertext environments; student's generally have different learning styles. A learning style is part of an individual's cognitive structure and refers to the pattern of behaviour and performance by which an individual approaches educational or learning experiences. The term learning style is used as a description of the a t t r i b u t e s and behaviours which determine an individuals preferred way of learning [12]. Individuals approach the learning of new ideas and concepts differently. Esichaikul et al. [13] advocate that authors need to take into account learning styles when designing hyperdocuments. For example, systems could include paths t h a t complement or accommodate the preferred learning style of a particular learner. The students were sequentially assigned to one of three t r e a t m e n t groups, each using a different t r e a t m e n t method (a lecture, a cognitively adapted hypertext or a static hypertext). Those students allocated to the lecture group were given a presentation and guided through some worked examples, lasting a total of one hour. Students in the cognitively adapted hypertext and the static
304 hypertext groups spent one hour learning as much of the information in the s y s t e m s as possible. Upon finishing t h e i r respective sessions, s t u d e n t s completed a test to ascertain the a m o u n t of material learned. The learning outcome was m e a s u r e d by a fourteen question post-test, which resulted in a single score. Questions were developed to test u n d e r s t a n d i n g of the key concepts and basic procedures. 5. R E S U L T S
Three research questions where addressed: (1) Does L e a r n i n g Style affect l e a r n i n g outcome? (2) Does the P r e s e n t a t i o n Method (Lecture, Cognitively Adapted Hypertext or Static Hypertext) affect learning outcome? (3) Are there interaction effects between learning styles and method of p r e s e n t a t i o n t h a t affect learning outcome? The dependent variable used in the study was an objective m e a s u r e m e n t of learning outcome. The independent variables include: (a) the learning style of the l e a r n e r and (b) the method of p r e s e n t a t i o n of the material. U n i v a r i a t e analyses of variance were used to test for significant differences in the means of the learning outcome scores between the different methods and the learning styles. Specifically, a two-factor ANOVA was used to investigate interactions effects between presentation method and learning styles. The learning outcome score means for each of the study design groups are presented in table 1, along with the number of subjects in each group and the standard deviation. Table 1. Learning Outcome by Learning Style and Presentation Method
Activists; means s.d. n
Reflectors; means s.d. n
Theorists; means s.d. n
Pragmatists; means s.d. n
Method Totals; means s.d. n
Lecture
Adaptive H~,pertext
Static H~cpertext
Learning Style Totals
18.90 6.74 10
17.67 3.85 12
14.50 4.03 10
17.06 5.15 32
19.11 2.57 9
16.27 3.20 11
14.10 5.39 11
16.32 4.36 31
19.00 1.73 3
17.75 1.71 4
12.00 3.92
16.00 4.05 11
17.50 4.51 4
21.00 6.24 3
16.50 3.42 4
18.10 4.59 11
18.77 4.63 26
17.50 3.74 30
14.28 4.47 29
16.79 4.63 85
4
,,
305 The differences among the means for the learning outcome scores across the three presentation methods were significant. (F2,82 = 8.2, p = 0.0006). Multiple range tests, indicated t h a t the static hypertext session m e a n is significantly different from the lecture and the cognitively adapted hypertext session at the 0.05 level. This separation of means is highlighted in figure 1. The differences among the mean scores between learning styles (F3,81 = 0.529, p = 0.664) were not significantly different and there was no significant i n t e r a c t i o n effect between learning style and presentation method on learning outcome.
21 4
20 19 18 17 6 -
5
=
[ 14. Lecture
13
Adaptive hypertext 12
D - Static h y p e r t e x t
Activists
Reflectors
Theorists
Pragmatists
F i g u r e 1. L e a r n i n g O u t c o m e P r o f i l e s for P r e s e n t a t i o n M e t h o d 6. C O N C L U S I O N This research a t t e m p t s to reduce the cognitive overhead and disorientation experienced by l e a r n e r s while p u r s i n g t h e i r i n s t r u c t i o n a l objectives in browsing a hyperdocument. Instructional cueing makes the conceptual model incorporated in the adaptive system explicit to the learner. This study reports the effects a conceptually adapted hypertext and an individuals learning style, have on learning outcomes with a hypertext system. The mean scores of those subjects who attended the lecture (mean = 18.8) and used the adaptive hypertext
306 (mean = 17.5) were significantly different from those who used the static hypertext (mean = 14.3). This would seem to indicate that students learn from an adaptive hypertext incorporating instructional cues, to the same level as students attending lectures, for the domain in question. Future research will determine how the methodology described herein generalises to other domains. The main result from the study is that there was no significant difference in learning outcome between the cognitively adapted hypertext system and the conventional lecture. There was however a significant difference between these two presentation methods and the static hypertext system. Hence adapting the hypertext document and providing instructional cues to the learner may reduce cognitive overhead and increase learning outcome. REFERENCES
1. 2. 3. 4. 5. 6.
7. 8. 9. 10. 11. 12. 13.
J. Nielsen, Hypertext and Hypermedia, San Diego, CA: Academic Press, 1990. W. Reader and N. Hammond, Computer-Based Tools to Support Learning from Hypertext: Concept Mapping Tools and Beyond, Computers and Education, Vol. 22, Part 1/2, pp 99-106 (1994). C. Tompsett, Knowledge-Based Support for Hypertext, 5th International Expert Systems Conference, Learned Information: Woodside: pp 31-43, (1989). J. Conklin, Hypertext: An Introduction and Survey, IEEE Computer, Vol. 20, Part 9, pp 17-41. (1987). F. Halasz, Reflections on NoteCards: Seven Issues for the Next Generation of Hypermedia Systems, Communications of the ACM, Vol. 31, Part 7, pp 836-852 (1988). T. Mayes, M. Kibby and T. Anderson, Signposts for Conceptual Orientation: Some Requirements for Learning from Hypertext, In R. McAleese and C. Green (Eds.), Hypertext: State of the Art, Intellect, pp 121129, 1990. T. Sokolnicki, Towards Knowledge-Based Tutors: A Survey and Appraisal of Intelligent Tutoring Systems, Knowledge Engineering Review, Vol. 6, Part 2, pp 54-95 (1991). T. Mayes, M. Kibby and H. Watson, StrathTutor: The Development and Evaluation of a Learning-by-Browsing System on the Macintosh, Computers and Education, Vol. 12, Part 1, pp 221-229 (1988). N. Hammond, Teaching with Hypermedia: Problems and Prospects. In H. Brown (Ed.) Hypermedia/Hypertext and Object-Oriented Databases. Chapman and Hall, pp 107-124, 1991. M. Merrill, R. Tennyson and L. Posey, Teaching Concepts: An Instructional Design Guide (2nd Edition), Educational Technology Publications, Englewood Cliffs, New Jersey, 1992. T.Williams and B. Bainbridge, Rule Based Systems, In G. Ringland and D. Duce (Eds.), Approaches to Knowledge Representation: An Introduction, John Wiley and Sons, pp 101-115, 1988. P. Honeyand P. Mumford, The Manual of Learning Styles, Honey UK, 1992. V. Esichaikul, R. Smith and G. Madey, The Impact of Learning Styles o n Problem-Solving Performance in a Hypertext Environment, Hypermedia Vol. 6, Part 2, pp 101-110 ( 1994).