Dr. Kim H. Veltman

Dr. Kim H. Veltman has taught at the universities of Toronto, Carleton, Göttingen, Rome and Siena. For the past twenty two years he has worked mainly as a post-doctoral fellow, focussing on the history of perspective, Leonardo da Vinci, and developments in new media. For the past two decades he has been working on a System for Universal Media Searching (SUMS). A comprehensive curriculum vitae is available at http://www.web.net/~akfc.

1. Introduction

2. Visualising Connections in Conceptual Spaces

3. Seeing Invisible Differences

4. Comprehension and Prediction by Seeing Absence

5. Conclusions

This paper outlines strategies and methods for tackling the enormous challenges presented by an emerging Information Society in which the resources of libraries and museums are gradually being made available on-line in electronic form. It begins from

two fundamental premises: first, that the experience of libraries, museums, archives and similar institutions in organising, ordering, classing and accessing knowledge is an obvious point of departure for serious strategies of search and access. A second premise is that the methods used for presentation of knowledge in libraries offer valuable clues for a coherent access, interface and strategy, offering a key to a common look and feel for all our activities, be it creating, classing, publishing or accessing.

Following from these premises is a new approach to the traditions of knowledge collection, organisation and retrieval. At one end of the spectrum there has been a dream that everything could be collected in one centralized institution. This inspired the Library of Alexandria, the British Museum and a host of other efforts. At the other end of the spectrum there has been an assumption that everything could be decentralised in a completely distributed system. Our claim is that neither of these extremes can work, which means that a new model is called for: a centralised repository of meta-data, a digital reference room which is effectively a cumulative collection of all existing reference sections in libraries and museums.

Libraries, in particular their reference sections, have constructed enormous lists of materials, notably their (now electronic) catalogues of author names, subject headings, keywords, titles, place names, call numbers and the like. These catalogues offer presentation methods to gain better access to their own collections of books, and other materials. At the same time, these catalogues are effectively authority files with standard and variant versions of authors’ names, place names etc. and as such can be used as instruments for navigating through materials elsewhere.

From a global point of view reference rooms in libraries contain mainly five kinds of materials, namely, 1) terms (classification systems, subject headings, indexes to catalogues); 2) definitions (dictionaries, etymologies); 3) explanations (encyclopaedias); 4) titles (library catalogues, book catalogues, bibliographies); 5) partial contents (abstracts, reviews, citation indexes). All of these are pointers to the books in the rest of the library, or 6) full contents which can conveniently be divided into another four classes, 7) internal analyses (when the work is being studied in its own right); 8) external analyses (when it is being compared or contrasted with other works); 9) restorations (when the work has been been altered and thus has built into it the interpretations of the restorer) and 10) reconstructions (when the degree of interpretation is accordingly larger). From this global point of view the first six of these categories are objective, while the last four (6-10) are increasingly subjective. Hence the physical arrangement of major libraries serves as a starting point for the conceptual system; the presentation system of libraries offers another key to access into its electronic version. The heritage of experience in organising the known provinces of knowledge, offers the departure points into its unknown lands.

The lecture included a demonstration of an existing prototype and outlined a blueprint for its further development. The prototype is called a System for Universal Media Searching (SUMS) and was chosen as one of 19 projects to represent Canada at the G7 Information Society Exhibition in Brussels (February 1995), at the G7 Summit in Halifax (June 1995); as part of G7 pilot project 5 (Multimedia Access to World Cultural Heritage) at the Information Society and Developing Countries (ISAD) Conference in Midrand (May 1996); as an invited signatory to the Memorandum of Understanding for Multimedia Access to Europe’s Cultural Heritage and as a member of the European Commission’s MOSAIC project, designated as the museums section of the Trans European Networks (TEN) Project.

The basic interface relies on three elements 1) six familiar questions: Who?, What?, Where?, When?, How?, Why?; 2) lists of choices which begin with ten basic notions: access, learning, levels (of knowledge), media, quality, quantity, questions, space, time, tools; 3) maps as a way of navigating spatially. The main body of the longer paper outlines the role of each of these, focussing on the power of strategic use of simple questions, the significance of purpose as orientation, the use of quality and quantity as means of refining queries; the role of maps and geographical modes, and the use of integrating tools. The System for Universal Multimedia Access (SUMMA) builds upon this approach by using the notion of a digital reference room as a key to universal access (see Appendix 1).

Traditional presentation and access methods have been two-dimensional. A comprehensive system must incorporate these traditions and develop means of moving seamlessly from two- to three-dimensional spaces. The latter part of the full paper explores some of the potentials of these emerging technologies in order to visualise connections in conceptual spaces and for purposes of comprehension and prediction by seeing absence as well as achievement.

Library cards present their information on a two dimensional surface and electronic catalogues initially followed this format. This was partly a reflection of the limitations of computing power which are rapidly being superceded. Accordingly a number of individuals are assuming that two-dimensional navigation will soon be discarded as a thing of the past and that we shall use exclusively three-dimensional spaces for all navigation. For example, Dr. Henry Lieberman (MIT) is exploring the use of very large navigation spaces, with new techniques which allow "zooming and panning in multiple translucent layers." Carnegie Mellon University has a special Visualization and Intelligent Interfaces Group. Silicon Graphics Inc. (SGI) foresees the use of landscapes. Dr. Stuart Card (Xerox PARC) and his team have been working on a series of tools for visualizing retrieved information using techniques such as a galaxy representation, spiral calendar, perspective wall, document lens and cone tree . There is an analogous project at the Gesellschaft für Mathematik und Datenverarbeitung (GMD) in Darmstadt called Lyberworld

Historically the advent of three dimensional perspective did not lead artists to abandon entirely two-dimensional (re-)presentations. There were many cases such as cityscapes where three dimensions were very useful; others where two-dimensional solutions remained a viable and even preferable alternative. Three-dimensional navigation spaces are particularly valuable for contextualising knowledge. A two-dimensional title or frontispiece of a book tells us nothing about its size. A three-dimensional scale rendering helps us to recognize at once whether it is a pocket sized octavo or a folio sized book: a slender pamphlet or a hefty tome. Hence, having chosen a title one will go to a visual image (reconstruction) of the book; see, via the map function, where it appears on the shelf of a library, do virtual browsing of the titles in its vicinity or wander elsewhere in the virtual stacks. In the case of more complex systems such as the London Underground it is useful to move progressively from a two-dimensional schematic simplification of the routes to a realistic three-dimensional rendering of the complete system, station by station.

The third dimension has many uses beyond producing such electronic copies of the physical world. Pioneers of virtual reality such as Tom Furness III, when they were designing virtual cockpits realised that pilots were getting too much information as they flew at more than twice the speed of sound: the challenge was to decrease the amount of information, to abstract from the myriad details of everyday vision in order to recognise key elements of the air- and land-scape such as enemy planes and tanks.

These principles are equally important in knowledge organisation and navigation. A library catalogue gives the works of an author. Each catalogue entry shows under how many fields a given article of book is classed. Adding these fields together leads to an alphabetical list of that author’s intellectual activities. In electronic form producing such a list is theoretically simple. What is needed, however, is a conceptual map. To what extent did an author work as a generalist in large subject fields and to what extent as a specialist? This lends itself to three dimensions. Author A is in one plane and the subject headings of their works are on other planes. These are aligned to relative positions in classification systems such that one can see at a glance to what extent this person was a generalist or a specialist (figure 1). This principle can be extended in comparing the activities of two authors.

This approach can in turn be generalised for purposes of understanding better the contributions of a group, a learned society or even a whole culture. Persons have a series of networks: correspondence, telephone and e-mail which help us to visualise the complexities of remarkable individuals be it in the world of the mind, politics or business.

General level

of Classification

More detailed level

of Classification

Author

Figure 1. Visualisation of an author’s activities whose specialist activities touch on four fields (three of which are closely related) and whose more generalist activities are limited to one major field. Further layers could be added to show how the same concepts recur in different places in various classification systems.

We could take the members of a learned society or some other group and trace how many layers in the classification system their work entailed and then study how this changed over time. Are the trends towards specialisation in medicine closely parallel to those in science or are there different patterns of development? Alternatively by focussing on a given plane of specialization we could trace which authors contributed to this plane, study what they had in common in order to understand better which individuals, networks of friends and which groups played fundamental roles in the opening of new fields. Navigation provides virtual equivalents of journies in the physical world. It is also a means of seeing new patterns in the conceptual world through systematic abstraction from everyday details in order to perceive new trends.

Contextualising also entails seeing relations between one subject and another. When we are studying a subject, we typically want to know about related subjects. In the past we went to a library catalogue, found a title and saw the related topics at the bottom of the card. Electronic versions thereof exist. Recent software such as Apple’s Hotsauce allows us to go from a traditional two-dimensional list of terms, choose one, and then see all the related topics arranged around it. These related subjects evolve with time, so with the help of a simple time scale we can watch evolution of a field’s connections with other subjects.

This notion of planes can be extended to see further patterns. Plane one can list all the known problems or potential research areas in a given area of science. Plane two lists which subset of these problems is presently being studied. Plane three shows which problems have been solved, or rather have some solutions in the form of inventions, patents, trademarks and products. Plane four lists a further subset for which solutions are predicted or which have hypotheses for their solution (figure 2).

Plane 4

Problems Predicted

Plane 3

Problems Solved

Plane 2

Problems Financed

Plane 1

Problems Identified

Figure 2. Using spatial arrangements of concepts to map problems identified and to visualise which subsets thereof were financed as research projects, which were solved in the sense of producing patents, inventions and products and which led to new predictions in the form of hypotheses and projections.

Such comparative views will help scientists and decision-makers alike to understand more clearly trends in rapidly developing fields. Such matrices of problems can in turn be submitted to problem structuring methodologies whereby technical, practical and emancipatory dimensions are submitted to frameworks in order to discern where they fit into a Methodology Location Matrix.

During the Renaissance the discovery of linear perspective brought new skill in visualising the physical world, but it began by illustrating episodes from the lives of saints which none of the artists had witnessed personally. Hence it helped simultaneously in expanding the horizons of the visible world of nature and the invisible world of the mind. This dual development continues to our day. Three-dimensional visualisations, especially using virtual reality are helping to illustrate both the visible and invisible, and introduce many new possibilities.

If, for instance we take the Library of Congress classification, as above, and link each layer in its hierarchy with a different layer, then we arrive at a truncated pyramidal shape beginning with twenty initial topics at the top and increasing to many thousands as we descend. Say we are interested in total publications in the Library of Congress. At the top level, these publications can be linked to each of the twenty basic fields, such that each major subject is represented proportionately as a square or circle. We can thus see at a glance to what extent the number of books on science is greater than those in fine arts.

By going down a level in the hierarchy we could see how those figures break down, e.g. to what extent there are more books on physics than chemistry or conversely. At another level we would see whether and if so to what extent astro-physics has more publications than bio-physics or quantum physics. We are thus able to see patterns in knowledge which we could not see simply by looking at the shelves, although shelves can provide some hints that one topic has more books than another.

A slightly more refined version would link this approach to book catalogues such that we can trace how these trends in publications have changed over time. From a global point of view we could witness the rise of the social sciences in the nineteenth century. At a greater level of detail we could see the rise of psychology as a field. This same approach could also be applied to usage patterns as studied by scholars in reception theory. In future usage patterns by on-line readers will become important.

In our quest to see significant patterns it will sometimes prove useful to have agents examine trends and draw our attention only to cases where there are considerable changes. This will be another way to discover emerging subjects. Hence, instead of trying to keep everything visible at all times, the system only brings to our attention those areas where trouble could occur: an electronic equivalent of preventative medicine. Such strategies will no doubt be aided by the further development of sensory transducers whereby significant changes in heat within the system would also be rendered visible. Seeing the otherwise invisible is a key to navigating remotely through complex environments.

In the early days of the scientific revolution there was great emphasis on the importance of inductive as opposed to deductive research, which entailed an emphasis on experience, experiment, often on a trial and error basis. As scientists gained a better understanding of the field to the extent that they were able to create physical and conceptual maps of their objects of study, it became possible to deduce what they had not yet observed. Once there was a periodic table, for example, chemists found that the known chemicals helped them to chart the positions of as yet unknown compounds. Once we have a matrix we can see where there is activity and where activity is missing. By now, chemistry is expanding with enormous speed. It is estimated that every week over 14,000 new chemical combinations are discovered. As in the case of pilots flying at twice the speed of sound we need methods for abstraction from the day to day details, new ways of seeing patterns. Access, searching and navigation are not just about seeing what we can find, but also about strategies such that we see the appropriate subsets at any given time.

To see a bigger picture we need to be able to see how the tiny details fit into the larger categories of human endeavour so that we can discern larger patterns. Roget had six basic classes. Dewey had ten: 0) generalities; 1) philosophy and related disciplines; 2) religion; 3) social science; 4) language; 5) pure sciences; 6) technology; 7) the arts; 8) literature; 9) general geography and history. The Library of Congress has twenty such fundamental classes. Beneath these universal headings are many layers of subordinate categories hierarchically arranged. If we treat each of these layers as a plane, and have a way of moving seamlessly from one plane to the next, then operations performed at one level can be seen at various levels of abstraction.

Suppose, for example, that we have been searching for Renaissance publications by Florentine authors. Moving up to the highest level we can see on which fields they wrote: religion, science, art and so on. Moving back down a few levels we can identify which particular branches of science and art concerned them most. Going back to the top level we can also see that there were many topics which they did not discuss. The Renaissance view was supposedly universal in its spirit. In practice, it often had distinct limitations. If we have access to multiple classification systems, then we can see how these patterns change as we look at them say, through the categories of Duke August and Leibniz at Wolfenbüttel or through the categories of Ranganathan’s system.

Such book catalogues and classification systems are the most important efforts at bringing order to the world in terms of subjects. But subjects in isolation are still only somewhat ordered information. Meaning which brings knowledge and wisdom requires more, namely a systematic ordering of these subjects in terms of their logical and ontological relations. Efforts in this direction go back at least to the I Ching. Aristotle, Thomas Aquinas, Ramus, Francis Bacon and Roget were among the many contributors to this tradition. In our generation, Dr. Dahlberg presented these fundamental categories in a systematic matrix. More recently these have been adapted by Anthony Judge into a matrix of nine columns and nine levels (figure 6), which generates a chart of 99 subjects. These range from fundamental sciences (00), astronomy (01) and earth (02) to freedom, liberation (97) and oneness, universality (99). Anthony Judge is using this as an "experimental subject configuration for the exploration of interdisciplinary relationships between organizations, problems, strategies, values and human development".

Heiner Benking, another speaker at the German chapter of the ISKO conference builds upon the same framework in his conceptual superstructure or cognitive Panorama Bridge, which is the basis of his Rubik’s Zauberwürfel [Cube of Ecology or Magic Cube]. He argues that one can use planes in order to see patterns in thought. These planes can include continua between the animate and the inanimate on one axis and between micro-, macro- and meso-scales on another axis. The planes can be used to see relations among different actions, options and strategies. They can be used to see different levels of abstraction and compare different viewpoints at a conceptual as well as at a perceptual level.

This paper was prepared for a lecture at a recent meeting of the German Chapter of ISKO devoted to Wissensorganisation mit Multimedialen Techniken [Knowledge Organization with Multimedia Techniques], (Berlin, October 1997), and is the abridged version of a longer paper which will appear in the international ISKO journal, Knowledge Organization.

Recent advances in technology assume a separation of content and presentation with respect to data structures. In terms of access, however, there are important reasons for relating content and presentation (different views, perspectives). The paper outlines some fundamental concepts underlying a prototype for a System for Universal Media Searching (SUMS), namely, learning filters, and knowledge contexts, levels of knowledge, questions as strategy: purpose as orientation; media choices, quality, quantity, questions, space using maps and projections; multi-temporal views and integrating tools. It foresees how such a system, linked with the equivalent of a digital reference room, will provide the basis for a System for Universal Multimedia Access (SUMMA). The latter part of the paper addresses recent developments in three-dimensional interfaces. It claims that these are particularly suited for certain tasks such as visualising connections in conceptual spaces; seeing invisible differences and comprehension and prediction by seeing absence. It suggests also some ways in which two- and three-dimensional interfaces can be used in complementary ways.