The main characteristics of hypermedia is that information consisting of different media elements are stored in atomic nodes and the relationship between these nodes enables the multidimensional association of information (Tomek et al 1991). Information contained in the nodes is made available to the user via links corresponding to embedded pointers. The presence of a link in the document is marked by some system-dependent marker such as an icon, or by highlighting.

When a hypermedia link is activated, the system extracts its destination node from the hyperbase and presents it to the user. Information in a typical document may consist of text, graphics, animation, digitised sound etc. After having examined the node, the user can then follow this node’s own links for further information and so on to any desired depth.

It would seem that at present there is an increasing desire to present education in a non-linear manner, and also to supply text, graphics, audio, video, and animation under computer control through various levels of interconnectedness and interaction. Tele-education can now exploit the power of new technology and incorporate video clips, voice annotations attached to reference text, reference to textual information from digitised video, and so on. Another advantage such systems offer to the user are the ability to interact with information in a non-sequential way, the user selecting their own path and navigating it at their own pace, acquiring knowledge by discovery and exploration.

Such a wealth of information in a highly interconnected network requires that the information be presented in a coherent, easy to assimilate way. Otherwise, cognitive and navigational overheads may arise when users interact with such systems. Reaching total familiarity with a concept to be learned is sometimes made difficult by the need to pay attention to the complex mechanics of the system itself as well as the richness of the information stored in it. Attention is thus diverted from the contents to the mechanisms of the system and the need to deal with the complexity of available behaviours.

Metaphors may be seen as a tool used to overcome such problems in hypermedia applications. Metaphors are used to simplify complex unfamiliar situations through embedding familiar objects and experiences. The user is presented with a familiar environment based around metaphors selected and represented so that the users immediately recognise what they are and what their functions are in the system. Metaphors can therefore reduce complexity and at the same time maximise power and flexibility (Davis et al 1991). For example, an unfamiliar real life scene may be presented overlaid with familiar computer-generated objects.

Nelson first coined the word ‘hypertext’ as a means of describing a semantic network of knowledge (Nelson 1965). Hypermedia takes this further, in that information consisting of different media elements is stored in atomic nodes and the relationship between these nodes enables the multidimensional association of information (Tomek et al 1991). There is an increasing interest in offering education, particularly tele-education, in a non-linear manner, as well as providing an integration of the associated materials in a variety of media types, text, graphics, audio, video.

Such systems allow learners to interact with information in the non-linear way psychologists consider the brain operates. Within an educational setting, the interconnected domains of knowledge allow the student to acquire knowledge through discovery and exploration, at the same time offering the learner freedom to choose the path they desire at their own pace. The learner can learn by discovery and personal experience rather than being a receptacle for the transmitted knowledge of a teacher; they can browse and intuit rather than being limited by the educational straightjackets of task oriented, programmed teaching.

Hypermedia-based systems have been in use in education in a wide variety of subject areas (Tomek et al 1991). From an educational standpoint the principal attraction of hypermedia is that it lends itself naturally to non-sequential educational approaches since it encourages the free association characteristics of human thought. It enables the learner to choose his or her own direction while browsing through an electronic book, moving from one knowledge domain to another in a process of information seeking and exploration (Hardman 1994).

Hypermedia is both an object and a process of education. It is the object in that it contains at the node level the content information to be presented to the student; it is the process in that it also contains the dynamic links that lead from one node to another. The links support both associative learning and learning by exploration and discovery (Norman 1994).

However, due to the very nature of the data structure of such systems, where a maze of information is stored at a node level and there are complex spatial relationship between these nodes, designers of hypermedia systems are faced with the problems of cognitive overloading and its associated problem of navigational overhead Tomek et al 1991). For example, students frequently become ‘lost in hyperspace’, so bemused by the wealth and choice of information on offer that they are unable to perceive the overall structure and progress productively through the material (Conklin 1987).

Two broad approaches to overcoming these problems can be identified. One is to make the hypermedia systems capable of adapting to their users so that, for example, different users are offered different numbers and types of hypermedia links. A second approach is to make the same links available to all users, but seek to assist the user through interface aids in the form of metaphors. We are currently conducting research concerning both of these approaches, with a view to offering increasingly powerful student-based learning opportunities. This paper is concerned with the second approach; an overview of our research concerning the former approach can be seen in Mullier et al (1998) and Moore et al (1997).

When a user begins the process of navigating the information space he may find that the target domain to which it is applied is much richer and more complex than expected. Hypermedia can seem to users to be more dimensional and more dynamic than expected (Norman 1994). Thus, cognitive overloading and navigational overhead may cause many problems to users interacting with such systems (Tomek et al 1991, Davis et al 1994). The main problems have been identified as:

- inability to obtain a useful measure of the amount of available information and the structure of that information

- difficulty in locating the desired information.

- inability to retrace the path to a known region of the hyperspace

- difficulty in using an unfamiliar interface

The cause of cognitive overloading is mainly the inability to create and provide a coherent structure for the user to obtain a rapid overview of the system functionalities. One of the main aims of a user in an educational environment is to explore a concept. In order to achieve this he should be able to build a mental model from what is presented to him. The ability to build a mental model depends on presenting a coherent structure to the user. The construction process is facilitated when the document is set out in a well-defined structure and provided with rhetorical cues or links reflecting these structural properties (Charny 1987, Carrol et al 1988).

Users should be able readily to interact and navigate the information space. Navigational overhead arises from the process of interacting with the information, in searching and browsing for what is required whilst lacking the ability to maintain a view of the links and the nodes that are not directly linked to the active node (Tomek et al 1991, Charny 1987, Moor & Hanky 1989). Empirical research shows that these additional activities may ‘interrupt the train of thought’ (Gordon, Gustavel, Moore and Hanky, 1989; Monk, Walsh and Dix, 1988). This may create a navigational overhead; navigation in an educational environment should thus be convenient and require as little effort as possible in order to minimise this overhead. One possible way to achieve this is through the mechanism of metaphors.

The development of WIMPS and direct manipulation interfaces has enabled users to interact with metaphorical interfaces in which they manipulate objects such as icons, windows, menus and scrollbars to control the underlying system. Metaphors have developed further so that in well-designed interfaces these objects are selected and represented in such a way that users intuitively recognise what they represent and how they function in the system from their prior knowledge and experience.

The use of metaphors may help overcome such problems where lengthy descriptions or technical jargon are generally needed for the learner, particularly a novice, to acquire a new skill or new knowledge. A metaphor is used in such situations by making reference to something with which the learner is familiar. They can be very powerful devices to communicate even a complex, structured set of properties in shorthand that is easily understood. For example, if a person describes his job as being like a ‘prison’ he communicates all properties of the superordinate category ‘prison’ with just that statement – the job is presumably confining, unpleasant, difficult to escape from, and so on. It is not possible to list all of these properties exhaustively, and the use of this verbal metaphor is more efficient and more precise than any partial listing of the properties of the superordinate ‘prison’. In other words, if the attribution of all those properties is the purpose of the communication, then the appropriate communicative form is the metaphor (Oratory 1985).

This becomes readily apparent in the sciences where it is quite common to use metaphors for conceptualising abstract concepts in terms of the apprehensible (Gentler, 1982). Hence metaphors could be used to help users to construct an appropriate cognitive representation of computing systems in general (Carol and Thomas 1982) and user interfaces in particular (Carol, Mack and Kellogg 1990, Ericson, 1990).

Some well-known examples of presentation metaphors are the Desktop metaphor (the computer screen presents a number of objects that are normally found on a conventional desk), the Book metaphor (in which users see an open book on the screen with a table of contents and pages to turn showing how much of the book has been seen), Guided Tours (which enable students to 'get on a bus' for a guided tour through the information space), and graphically represented 'Rooms' (where 'opening a door' maps onto a workplace tailored for some overall task). Other metaphors include Library, Map, Chalkboard, and Physical World.

Metaphors can also be defined as analogical models to something existing outside the domain to which they are applied. Thus the knowledge in one already-familiar domain may be directly applicable to another, less familiar domain. Such metaphors may be useful in user interfaces so that the user can rapidly adopt a correct mental model of how the system works. (Nadeau 1996, Glowella 1995). Hypermedia requires structural metaphors because of the inherent problem in the hyperstructures of ‘getting lost in

hyperspace’. The structure of the metaphor may then lend itself to impose additional structure and landmarks on the hypermedia application domain. Thus, various structures such as a group, a hierarchy or a network of the multimedia information could become more familiar when presented as a building, room, book, city or a landscape (Vaananen 1993, Diesberger 1993) or new metaphors such as gardens, flowers, trees as appropriate for the application to be presented could be

generated.

The dynamic guidance found in some of the above metaphors guides a user to the information available in nodes that are not directly linked to the active node. This may be used to enhance and facilitate the different navigational tasks of searching, browsing and orientation. Metaphorical tools may be used for this purpose. Tools such as guided tours, narrated tours, guides, fisheye views, trails, mirrors and videoclips can enhance and facilitate searching, browsing and orientation.

Guided tours allow the user to make use of a virtual bus or bicycle to take a tour around the information space. A video clip can be activated to obtain an overall view of the available information at any given point for reorientation. Another tool is a graphical representation of a tutor who is present as a guide at all times to advise the user about the best route to take and about the information available at any given content node (search and browsing). These different tools can have different effects on the various navigational tasks (Tomek et al 1991, Davis et al 1991, Garzotto et al 1996, Allison and Hammond 1988).

The ‘guided discovery approach’ is a valued tool in a learning environment to support controlled guidance without giving the student complete control of the navigation (Laurillard 1995). The lessons using dynamic guidance could contribute towards this approach.

Although metaphors are widely accepted as being of value there is little hard experimental evidence associated with their use. Thus, although anecdotally they may seem to help the learner, they may well be not operating as efficiently as they could or, indeed, a particular metaphor may be inappropriate for certain categories of user.of user. Metaphors in large hypermedia systems have been studied by Davis et al (1991), and the cognitive representation of metaphors in computing by Carrol et al (1988); Smith et al (1995) have provided a reflection on the design of interface metaphors.

Nevertheless, the work so far undertaken to study the use and characteristics of metaphors has been limited. The aimThe aim of the current research project is therefore to explore the nature of metaphors and the learner's perception of interacting with them. A series of experiments is planned around prototype hypermedia systems employing a variety of metaphors. Different concrete metaphors such as those described above will be used to explore their potential with respect to conveying the structure of, for example, a classroom, a university, a city or a book. An appropriate authoring tool will be used to generate inter-linked display frames in a hypermedia structure. Each frame will include graphical and textual information, animation, sound and time-delayed components.

The expectation is that an analysis of the results will lead to an evaluation of the effectiveness and efficiency of various metaphors with respect to overcoming the problems associated with cognitive overloading and navigational overhead. A follow up series of experiments will then create prototypes using a variety of dynamic tools for different navigational tasks. The results of these experiments will be analysed to evaluate the effectiveness and efficiency of such tools to facilitate navigation.

In conclusion, it seems that hypermedia is increasingly being perceived as a suitable mechanism for delivery of educational courses, particularly those offered through tele-education. However, the supposed benefits to on-line distance learners will only be fully realised when the problems associated with hypermedia are addressed and resolved. It is the contention of this paper that a better understanding of metaphors could lead to a significant improvement in the current interfaces to hypermedia educational systems and indeed to other user interfaces as well. These findings, in the form of recommendations and guidelines, will unfold with the research study and will feature in a future paper; their value will be in allowing the design of more effective distance and tele-education.

L. Allison, N. Hammond (1988). Travels around a learning support environment; Rambling, Orienteering or touring: CHI 88, 269 - 273.

J. Conklin (1987). Hypertext: An introduction and survey: IEEE computer, 20, 17 - 41.

J. Carrol, R. Mack, W. Kellog (1988). Interface metaphors and user interface design: In handbook of Human Computer Interaction (M. Hallender, ed), Elsvier publishers, Holland, 67 - 85.

J.Carrol, J. Thomas (1982). Metaphors and the cognitive representation of computing. In handbook of Human Computer Interaction (M. Hallender, ed), Elsvier publishers, Holland, 95 - 101.

G. Charny (1987) Comprehension of non linear text: The role of discourse cues and reading strategies: Proceedings of the first ACM Conference in Hypertext, 109 -119.

G. Davis. H. Maurer, J. Preece (1991). Presentation metaphors for a very large Hypermedia systems; Journal of Microcomputer Applications 14, 105 - 116.

N. Ford, R. Ford (1992). Learning strategies in an ' ideal' computer-based learning environment: British Journal of Educational Technology 23, 195 - 211.

S.Gordon, J. Gustavel, J. Moor and J. Hanky (1988). The effects of Hypertext on reader knowledge representatation: Proceedings of the Human factor society 32nd Annual Meeting, 296 - 300.

F. Garzot. L.. Mainet. P. Paolin (1996). Navigation in hypermedia applications, modelling and semantics; Journal of organisational Computing and Electronic Commerce 6, 211 - 237.

U. Glowella ( 1995). Metaphors for hypermedia interfaces; Designing user interfaces for Hypermedia 1, 55 - 57.

L. Hardman (1995). Experience in authoring Hypermedia; Creating a better presentation: Designing user interfaces for Hypermedia 1, 18 - 28.

D. Laurilard ( 1995). Multimedia and the changing experience of the learner; British Journal of Educational Technology 26, 179 - 189.

A. Monk, P. Walsh, A. Dix (1988). A comparrison of hypertext, scrolling and folding as mechanism for program browsing: People and Computers, Cambridge University Press.

D J Moore, D J Hobbs, D Mullier, C Bell (1997). Interaction Paradigms with Educational Hypermedia, in Proceedings of EUROMICRO conference, Budapest

D. Mullier, D J Hobbs, D J Moore (1998). A Web-based Intelligent Tutoring System; Proceedings of NETIES international conference, Leeds, UK

K. Norman (1994). Navigating the educational space with hypercourseware ; Hypermedia 6, 35 - 59.

J.Nelson (1965). A file structure for the complex; Proceedings ACM National Conference 84-100.

D. Nadeau (1996). User interface metaphor in virtual reality using RML: Behaviour Research methods, Instruments and Computers 28, 170 - 173.

I. Tomek, S.Khan, T. Muldner, M. Nasser, G. Novak (1991). Hypermedia- Introduction and Survey; Journal of Microcomputer Applications 14, 63 - 103.

M.Smith, B. Anderson, R. Knot, J.Alty (1995). Human Computer Interaction: Interact, 339 - 345.

K. Vanannan, J. Diesberger (1995). Metaphor based user interfaces for hyperspaces: Designing user interfaces for Hypermedia 1, 68 - 77.