macf

THE CAMBRIDGE EXPERIMENTAL VIDEODISC PROJECT

Alan Macfarlane Julian Jacobs Sarah Harrison Martin Porter

(Department of Social Anthropology, Cambridge)

BICA Issue No. 5: February 1987

In June 1983 the Department of Social Anthropology at Cambridge and the Audio-Visual Aids Unit of the University of London collaborated to produce one of the first academic videodiscs. Some members of the Department helped to assemble 10,000 still frames of an anthropological and academic nature which were re-photographed and transferred to the disc. Since then the interest in the potentials of videodisc has increased greatly. This is particularly so since BBC Domesday Disc, to be launched in 1986, attempts to provide a massive `scrapbook' of pictures, films, maps, texts and statistics about Britain nine hundred years after Domesday. A group in the Department of Social Anthropology in association with the Cambridge University Audio-Visual Aids Unit, directed by Martin Gienke, has decided to make an experimental disc of limited size to explore the potentials of videodisc in teaching and research. For those not familiar with the technology, a videodisc is an object like a gramophone record. Information is engraved on the surfaces which are then coated with plastic. The information can be read by a laser beam. This provides a stable and enduring store for a very large quantity of materials of different kinds. A standard Philips Laservision disc can store film which will play for 36 minutes, or hold 54,000 separate pictures per side. It can store 10\s-2\u10\d\s0 bits (approximately 300 megabytes) of digital information per side. In practical terms this means that one could store the whole of the Encyclopedia Britannica, pictures and words, on about two thirds of one side of a videodisc. The Public Archives of Canada estimated that one digitally encoded disc could hold 40,000 pages of text, 5,000 photographs, twelve minutes of moving footage, eight hours of narrated film strip presented at two frames a minute and 1,000 microcomputer programs. A videodisc linked to a CD ROM (Compact Disc Read-Only Memory) storage disc adds another 600 megabytes of digital storage (i.e. six more Encyclopedia Britannicas).

A videodisc can hold almost all kinds of recordable information: photographs, slides, moving film, x-rays, round recordings, graphics, printed works, manuscripts. Recent discs can hold both analogue and digital material, the latter making it possible to to input data directly from a computer. A disc has a visual track, which can contain both pictures and teletext-like information, and two sound tracks. It is relatively cheap to copy a disc once the master has been made: on a long run of several thousands the unit cost drops to a few pounds . A videodisc linked to a micro-computer permits three kinds of work. The simplest is a `guided tour', like a tutorial, lecture or television programme: information is presented in a sequential or linear form, and the user is passive. Next, an `interactive' program using the techniques of branching, loops and multiple choice: it is like computer aided instruction, but with a full range of visual and sound materials as well as text and graphics. Thirdly the videodisc provides a huge archive, library or museum of materials through which a user can browse or stroll: a Borges-like world, with almost infinite paths and connexions.

Most conventional means of disseminating information succeed only if they are directed at a specific audience. Books, articles, lectures, films, exhibitions all aim at an audience with specific knowledge, interests and intelligence. Our planned videodisc, and videodiscs in general, are much less restricted in this matter of the audience: as Peter Armstrong, Head of the BBC Domesday Project, argued in a recent talk, they combine many of the qualities of book, television programme, computer database and large exhibition. The amount and variety of data which we can store on a videodisc, combined with powerful modern microcomputers, allows us to plan for different kinds of audience simultaneously.

Videodisc is durable and tough; it is an excellent archival medium for the storage of both visual and written material; the compactness saves money and space; computer indexing makes very large sets of material accessible quickly, and makes the medium an attractive research tool. A major obstacle to research in a number of disciplines is the dispersion and inaccessibility of materials. For instance, the materials for our videodisc, on the Indian tribal group known as the Nagas, is dispersed in archives, museums, libraries, private homes all over Europe and in America and India. No-one has been able to study these materials in depth because of the labour of assembling them, and the impossibility of finding one's way through the large quantities of material by hand even if they were assembled.

We are also interested in the use of videodisc in teaching and the dissemination of knowledge. A teacher or lecturer from primary school to University can use videodisc to prepare interesting courses combining visual, sound, graphical and textual materials. These can be either straightforward teaching courses, or can allow students to interact with the materials. Experiments in Japan, where educational videodiscs are now becoming a commonplace, suggest that they improve the quality of education and are of particular value in science subjects. In England early trials have suggested that children and teachers welcome videodisc with enthusiasm. We are particularly interested in possible applications in Museums. A frequent criticism is that Museum objects have been wrenched out of context and that they should be combined with other media \*- films, sound, detailed explanations. Videodisc perhaps allows us to meet these difficulties. The Geology Museum in London has recently made a start, and we are interested in working with anthropological museums, particularly those with collections of Naga objects.

We have decided to base our experimental disc on the Naga tribes of the Assam border, a people living in mountainous country roughly the size of Wales. Our period of study is from the earliest records up to Indian independence in 1947, when the population numbered about half a million. We chose the Nagas for a number of reasons. Given our limited resources of money and manpower, the task was possible. A videodisc, being a half visual medium, requires a subject which has visual interest. The Nagas excel in the visual arts and in their costumes, carving, dance and ritual they make their social and ideological system external and manifest. Furthermore the records of their culture and particularly of their relations with British power, are unusually rich. Partly this is due to their strategic position and character. Living on a border, in thick jungle and precipitous hills, renowned for their head-hunting and bravery, it took over a hundred years for a combination of missionary, trading, administrative and military pressures to `pacify' them. This gradual process produced an unusually rich set of records over a long time-span. This was the result of another factor, namely that a number of British and other observers, military men, district officers and anthropologists decided to document in as much detail as possible what they saw as a soon-to-be-shattered but splendid and beautiful tribal world.

The major figures included the first two Professors of Social Anthropology at Cambridge, T.C. Hodson and J.H. Hutton. In addition, J.P. Mills, Christoph von Fu\*:rer-Haimendorf, Ursula Graham Bower, W.G. Archer and others set out to record the Naga tribes with a thoroughness which has seldom been equalled. Published texts amount to the equivalent of over a dozen monographs, most of them very detailed ethnographies. Unpublished materials include over a hundred tour diaries by just one of the district officers, and numerous files of field notes, diaries, letters and other sources are emerging. We use methods ranging from OCR (Optical Character Recognition by computer) to ordinary text input to mainframe, for dealing with this material. There are well over six thousand Naga objects in European collections, perhaps many more, varying from the exotic and beautiful to the very humdrum. We have photographed some of the best collections in Oxford, Cambridge and London thanks to the kindness of the Museums. The collections of early still photographs are equally rich. Fu\*:rer-Haimendorf and Ursula Graham Bower alone have generously put at our disposal nearly five thousand magnificent photographs, and there are others dating back to the 1860s. We are developing ways to convert these early negatives and slides into the new medium. There is also a reasonable amount of moving 16mm film, about five hours in all. As well as this, there are extensive collections of maps, mission records and some sound archives.

The advent of videodisc, and the availability of very large caches of ethnographic and historical materials would not in themselves have been enough to make our project viable. The indexing and cataloguing power of the computer make it possible to use the medium of videodisc. The Naga disc will hold hundreds of thousands of `texts', whether paragraphs of writing, captions, photographs or clips from films. Both in order to write `guided tours' and also to browse through the information freely we need indexing systems that allow flexible and immediate access to large amounts of data. We faced similar problems with an earlier project when we constructed a database of records concerning the history of an English village, and developed a Database Management System (CODD) and a query language (CHIPS). On the basis of that work and more recent experience with the BBC Domesday we will probably opt for a dual strategy in computerised indexing.

One approach to the materials will be through an hierarchical branching structure of four to five levels. For example, a possible hierarchy would be:

If the user chooses `Culture', the next level might be:

Within each of these headings would be anything from three to twenty or so `level three' headings. Thus the user might move up and down the levels, seeing level one as volumes, level two as chapters, level three as sections of chapters which contain `paragraphs' of visual or textual or sound information. In fact the data will not be rigidly compartmentalised, since users can move laterally through the materials, making their own connexions. That is made possible by the second method of indexing, through key-words, allowing free-text retrieval.

There are two methods of retrieval at our disposal for this second indexing procedure, Boolean and free text. We hope to develop a sophisticated system by supplying both full `probabalistic' retrieval with inquiry expansion and relevance feedback, as well as full Boolean retrieval. The power of this system and the meaning of these terms is explained in the manual to MUSCAT (Museum Cataloguing system) developed by Martin Porter as a successor to his earlier systems GOS and GOSLING. Martin Porter is the computing advisor on this project and will be writing the file handling and retrieval system. We hope that this will be a relatively simple general cataloguing system usable with other videodiscs, possibly called DISCAT.

In this direct access approach the two systems of Boolean and free text search can be combined. The Boolean vocabulary will be fixed by us, and will break up the material in an organised way. Thus the user could ask for

Angami and Food and Photograph but not Rice

Angami \fIand\fP Food \fIand\fP Photograph \fIbut not\fP Rice

to get the photographs of foods other than rice which the Angami tribesmen eat. A free text inquiry, `use of grubs, worms and insects' would operate within that subset and give the user examples restricted to the use of these items as food among the Angamis. The free text need not always be supplied by the user. If the user says `Is there another object like this one?' then the free text description (i.e. caption) for the object on display supplies the free text to pursue the inquiry.

The retrieval system will be written to ensure that the machine dependent parts (screen operations and necessary secondary storage) are well separated from the rest, so that the system is as micro-independent as possible, and also as independent of whatever is special to the Nagas themselves. We attach considerable importance to giving the software an appropriate degree of generality and portability. Fortunately a suitable first approximation to the retrieval system we envisage is available in MUSCAT, which can be enhanced and made more general for this application.

We should end by stressing that this project is in a very early stage. This is just as sketch of what we are starting to do and the technology is changing fast.

Return to Contents page