That multimedia knowledge bases are informational tools approached by users seeking information, and that as such they should reflect the needs of users, is axiomatic. Interaction with multimedia knowledge bases is a communication process in which success is determined by an alignment of users' knowledge structures and system structures, that is what users know and how they structure what they know, and what the designer knows of user knowledge and its structure as it is reflected in system structure. With particular reference to the development of a multimedia knowledge base for community activists, this paper addresses the identification of knowledge structures as a critical factor in the conceptualisation and structure of such knowledge bases. It focuses on the application of an integrated process of concept analysis and discourse analysis in order to identify conceptual neighbourhoods and their interrelationships for the purpose of developing navigational pathways to enable users to actively search the knowledge bait.
The Green Ban Period generated a massive amount of material, only a small amount of which has been collected in libraries and other information agencies. Much of the surviving material still remains in the hands of the 1970s activists. It includes pamphlets, posters, newspaper cuttings, graffiti, diaries, development proposals, film, video, and audio recordings. Historical researchers, past and current community activists, journalists and Australian archivists in recent years have noted the absence of provision of access to these valuable materials for use in a range of fields of study - sociology, politics, journalism and conservation, as well as an information base for current community activists. Access is limited and somewhat haphazard. To date, some attempt has been made to bring these materials together, however, no index of the intellectual contents of these materials has been developed. By their very generation, the materials are already linked in complex ways in a wide network of inter-connections, each used to illuminate and understand the others. As the collection is in a variety of formats that can be, by their very nature, approached in different ways and applied to different purposes, it is recognised that a simple subject based index system may fail to reveal the richness and complexity of the pathways by which they might be accessed and interpreted by different groups of users.
Empirical research however contains remarkably little evidence of end users' participation in the development of retrieval systems that are responsive to their particular information needs. The development of retrieval systems based on the intellectual contents of materials has focused on the application of existing indexing languages and the construction and evaluation of indexes based on these, and automatically generated indexes, by their very nature, have not been developed with end user participation. This process does not capture the variable number of messages and meanings based on context, situation, cognitive style and intended use by users. The multi-dimensionality of central concepts represented in materials is not made explicit in the organisation of the index (Schamber, Eisenberg & Nilan, 1990).
The usefulness of these indexing systems has been questioned in recent literature, with particular criticism made of commercial retrieval systems that are not developed to service end users (Harman, 1992). A growing amount of attention is now being paid to user characteristics, both for the design of interfaces and for the organisation of the knowledge base. It is imperative that the development of such systems proceed from a strong user base if the full capabilities of interactive multimedia systems are to be exploited. In a recent report for the Australian Museum on interactivity, Nash claims: "the revolutionary characteristic of interactivity, when applied to audio visual media, is that computers have extended in quantum leaps the range of flexible real time choice by users of the sound/ images/ text that they are viewing and/or producing." This has ramifications for the construction of narrative pathways through an interactive audio visual/textual data base. "Internationally there is now a growing move to question the value of interactives which have advanced technical specifications but embody comparatively primitive understandings of how people relate to images and data . The formal structures of narrative will change to accommodate audience participation." (Nash, 1992, pp173-174.) This notion is central to the development of the project: rather than constructing a system based on hypothesising what a user might do, it would be built on what users have done in past situations and what they are doing in present situations.
The second assumption is that it is reasonable to regard the study of human-computer interaction as a form of communication. Borgman (1986) suggests that the more promising theoretical areas to be explored in explanation of the interaction are those of dyadic communication and of cognitive models brought to it. The transactions in this interaction are goal driven, to match a user's information needs with system output, and their successful outcome depends on the ability of the end user to form a mind picture of their own need and of the system. The interface is limited in its ability to respond to the user by the view the system designer has of users, the user model embedded in the interface.
Kuhlthau (1988) argues that the search process in information retrieval is a process of construction. The topic of the search is construed and re-construed in the communication process. The initiating query statement is usually in the form of a 'label' of few terms, which may be conceptually far from the underlying information need. This statement is a product of the knowledge, image, or model that the end users have of the texts represented in the system of the underlying principles of their organisation, and also of an extended information environment which includes their socialised knowledge of the human activities from which information is generated (Katzeff, 1989). This highlights the importance of states of knowledge in explaining an individual's motivation to use the system.
The concept analysis approach is founded on Ausubel's assimilation theory of cognitive learning. This theory asserts that concepts play a central role in the acquisition and use of knowledge, and that new knowledge and meaningful learning result when a person consciously and explicitly ties new knowledge to relevant concepts and propositions already possessed. It asserts thus that concepts, and propositions composed of concepts are the central elements in the structure of knowledge. Concept meanings are developed primarily in the extent that they are embedded in frameworks of propositions, and that the set of propositions a person has incorporating a given concept defines that person's idiosyncratic meaning for that concept. Concepts thus do not have fixed meanings, but rather meanings derived from the matrix of propositions in which they are embedded (Novak, Gowin & Johansen, 1983, p.626). At a practical level, this approach focuses on externalising knowledge structures through: (a) determining the global structure of the message, that is, establishing perceptions of the hierarchy of concepts, from superordinate to subordinate; and (b)identifying the texture of the message, that is establishing the various meaning of the relationships. It enables people to articulate central concepts, clarify their meaning, and establish how and why they are linked.
Halliday and Hasan (1985) present a framework for discourse analysis. They assert that the written and spoken word contain multiple layers of meaning based on a multiplicity of contexts. The thread which links these contexts and which helps to clarify meaning can be identified through an analytical process. This process seeks to identify the subject matter (field), the relationships and values (tenor), and the purpose behind the conversation or text (mode). This approach extends the context within which an individual establishes a hierarchy of concepts and their propositional relationships.
In the identification of field, one would expect a statement of the subject area and how that Subject area relates to other subject areas, and a description of what is happening. Tenor is concerned with the major actors and their relationships to each other and to the action or subject area. Mode reflects what the participants expect from the conversation or the text, for example, whether the text is intended to inform or persuade.
Research in artificial intelligence, cognitive psychology and linguistics suggests that thew analytical approaches to externalising structures of knowledge are also embedded in network models of semantic memory and schema theory.
Quillian (Ringland & Duce, 1988) proposes a model in which concepts, corresponding to particular senses of words or phrases, are presented as nodes in networks. Properties relating to concepts are represented as the labelled relational, two way links from nodes to other nodes. Links are differentiated by a form of weighting, according to how essential each is to the concept meaning. This enables a network of nodes to be linked. The complete meaning of any concept is the entire network emanating from the initial concept node. Links can be of different types and are thus nested or embedded to any degree to allow expression of any level of abstraction or specificity.
A schemata contains the network of interrelationships of the concept, that is, the "data structures for representing the generic concepts stored in memory" (Rumelhart 1984, p.162). Schema theory explains how knowledge is represented and how that representation facilitates its use. It provides an understanding of the structures which guide information processing in: directing attention; providing frameworks or scaffolding for preserving episodic information as a memory trace; and in providing frameworks for integrating new information, including the use of metaphors and analogies, as mapping between frameworks. It is thus a means of explaining what user knowledge, including contextual knowledge, is brought to interaction, as impetus in information seeking and as a basis for the ability to interact, and a way of explaining how user knowledge changes in interaction.
The methodology to be described below will allow knowledge engineers and systems designers to:
The community activists who participated in the development of the conceptual framework comprised people who were involved in action during the Grew Ban period, contemporary activists, authors of Green Ban materials held in the private collections, and people researching the period for a variety of reasons. This group thus enabled knowledge from both generators and end users of the materials to be represented.
To elicit the concept clusters, interviews based on the focus group model were held. The group approach was chosen because of the acknowledged solidarity of community activists in Sydney. (This solidarity was confirmed in a pilot test of the technique). Groups of eight to ten activists were asked to describe a specific situation when they needed information from the Green Ban period to enable them to progress with their current community activities. A small collection of materials was made available to the group. This consisted of photographs, newspaper cuttings, poster and private diary entries. They each selected a number of Green Ban materials that met their situational needs and undertook a concept analysis of these materials. Once everyone had completed their concept analysis, a structured discussion using a series of questions and prompts was held, and this was recorded. After the formal group discussions had ended, a number of the activists volunteered to continue the discussion and these were also recorded.
The central concepts identified by community activists are multi-dimensional. Dimensions such as source, time, space, situational environment, degree, scope. prior experience, motivation, and viewpoint are important contextually in shaping the portrayal of the concepts. The representation of these dimensions are seen by community activists as important in judging the usefulness of the documents to their particular needs. These dimensions will need to be made explicit in the design of the system.
The interpretation of these concepts within each of these dimensions generates a high level of specificity in the meaning of concepts. It is important that system designers place these concepts in appropriate contexts so that the matrix of meanings is correctly established. For example, whether a poster is viewed as propaganda or informing can only be established by linkages to contextual elements such as intention, key players, situation, and time.
From our comparison of the outcomes of community activists and indexers, it is evident that community activists speak a unique language which they appear to understand and use with ease to communicate with one another. Their conceptualisation of the material is not an indexer's conceptualisation of it. Essentially, the message is very clear: systems that are based on a conceptualisation derived from traditional subject analysis (or automatically generated indexes) will be of limited usefulness to community activists.
Boone, R. & Higgins, K. (1991). Hypertext/ hypermedia information presentation: Developing a hypercard template. Educational Technology, 21-30.
Borgman, C. (1983). The user's mental model of an information retrieval system: Effects on performance. Ann Arbor, Michigan: University Microfilms International.
Borgman, C. (1986). Human computer interaction with information retrieval systems: Understanding complex communication behaviour. Annual Review of Information Science and Technology, 19, 33-61.
Borgman, C. (1989). All users of information retrieval systems are not created equal: An exploration into individual differences. Information Processing & Management, 25(3), 237-251.
Dervin, B., Harlock, S., Atwood, R., & Garzona, C. (1980). The human side of information: An exploration in a health communication context. Communication Yearbook, 4, 591-608.
Dervin, B & Dewdney, P. (1986). Neutral questioning: a new approach to the reference interview. RQ, 25(4), 506- 513.
Dervin, B. & Nilan, M. (1986). Information needs and uses. Annual Review of Information Science and Technology, 21, 3-33.
Franklin, C (1989). Hypertext defined and applied. Online, May, 37-49
Frisse, M. & Cousins, S. (1992). Models for hypertext. Journal of the American Society for Information Science, 43(2), 183-191.
Gauch, S. (1992). Intelligent information retrieval: An introduction. Journal of the American Society for Information Science, 43(2), 175-182.
Halliday, M. & Hasan, R. (1985). Language, context and text: Aspects of language in a social-semiotic perspective. Melbourne: Deakin University.
Harman, D (1992). User-friendly systems instead of user-friendly front-ends. Journal of the American Society for Information Science, 43(2), 164-174.
Katzeff, C. (1990). System demands on mental models for a full text database. International Journal of Man-Machine Studies, 32, 483-509.
Knee, M & Atkinson, S. (1990). Hypertext and hypermedia: An annotated bibliography. New York: Greenwood Press.
Kuhlthau, C (1988). Developing a model of the library search process: Cognitive and affective aspects. RQ, 28, 232-242.
Lunin, L. & Harman, D. (199 1). Perspectives on ... human-computer interface: Introduction and overview. Journal of the American Society for Information Science, 43(2), 153-155.
Marchionini, G. (1992). Interfaces for end-user information seeking. Journal of the American Society for Information Science, 43(2), 156-163.
Mylonas, E. (1992). An interface to classical Greek civilisation. Journal of the American Society for Information Science, 43(2), 192-201.
Nash, C. (1992). Interactive media in museums: Looking backwards, forwards and sideways. Museum Management and Curatorship, 11, 17 1184.
Nielson, J. (1990). Hypertext and hypermedia. Boston: Academic Press.
Novak, J. & Gowin, D. (1984). Learning how to learn. Cambridge: University Press.
Novak, L, Gowin, D. & Johansen, G. (1983). The use of concept mapping and knowledge vee mapping with junior high school science students. Science Education, 67(5), 625-645.
Ringland, G. & Duce, D. (Ed) (1988). Approaches to knowledge representation: An introduction. New York: John Wiley.
Rubin, H. & Rubin, I. (1992). Community organising and community development. New York: Macmillan.
Rumelhart, D. (1984). Schemata and the cognitive system. In Handbook of Social Cognition. VI. Hillsdale, NJ: ErIbaum Associates, 161-188.
Schamber, L., Eisenberg. M. & Nilan, M. (1990). A re-examination of relevance: Toward a dynamic, situational definition. Information Processing & Management, 26(6), 755-776.
Woodhead, N. (1991). Hypertext and hypermedia: Theory and applications. Wokingham, UK: Addison-Wesley.
|Contact: Ross J Todd, School of Information Studies, University of Technology, Sydney, Kuring-gai Campus, PO Box 222, Lindfield NSW 2070 Australia. Tel: 02 330 5518 Fax: 02 330 5528 Email: R.Todd@uts.edu.au
Please cite as: Todd, R. J., Parker, J. and Yerbury, H. (1994). Knowledge representation and multimedia knowledge base design: A methodology for alignment. In C. McBeath and R. Atkinson (Eds), Proceedings of the Second International Interactive Multimedia Symposium, 543-548. Perth, Western Australia, 23-28 January. Promaco Conventions. http://www.aset.org.au/confs/iims/1994/qz/todd.html