Indexes for the pixel age

As the world's pictures are digitised and powerful interests jockey for electronic rights, Charles Oppenheim explores the maze surrounding ownership of diverse works, Roy McKeown describes progress in image library cataloguing and Anne Ramsden recounts the Excalibur search engine's academic and business conquests.

Now that the technology for turning pictures into pixels has been more or less mastered, the question of indexing and documentation has come to the fore.

If we are dealing with stock shot photography which shows anonymous people doing generic things then a caption and a couple of keywords may well be enough to find the image when it is wanted. But a collection of art reproductions will need to include full information not only about each item but also about its creator, to place it in context and facilitate access to related works. Between these two extremes lies a whole range of possibilities.

Whatever is decided, it is crucial that the individual images are indexed consistently. A fixed keyword list might suffice where the collection is narrowly focused, modest in size and not being added to, but then again it may be that a full-blown thesaurus is required in order to deal with synonyms and the relationship of specific to generic terms. If the material is in a specialist field there may be a published thesaurus which can be adopted.

There is always a temptation to tailor the text elements of the database to what the existing data will support and use "common sense" rules but while this is possibly adequate for stand alone image banks problems start to arise when links into wider systems such as the institution's library are required. At that point recourse to standards is virtually unavoidable.

Considerable work has been done in this sphere by information professionals. The MARC standard has a long pedigree and forms the basis of most book cataloguing. The difficulty is that a comprehensive MARC record is expensive to generate. The concept of core records, less complete than a full MARC record but more informative than a minimum record, has been developed by way of a solution.

The Program for Co-operative Cataloguing has developed a core record for audiovisual materials and the Visual Resources Association a core record for visual materials. These have been mapped to MARC and to the Getty Information Institute's Categories for the Description of Works of Art .

It is perhaps unfortunate that development has been so heavily concentrated in the heritage sector but not surprising when the field is multilingual and even artists' names vary. (Pieter Bruegel the elder has 29 different variations according to the Union List of Artists' Names.) The rapid development of the Internet is beginning to break down this sectoral emphasis, with the expectation that all resources are potentially available to everyone.

Some sort of translation device is required so that systems can exchange information in a meaningful way. Z39.50 is an ANSI/ NISO standard which lays down a protocol enabling systems to communicate. It regulates the computer commands and responses involved in transactions and includes a template of the database fields known to it. Systems which comply with the standard can then equate the fields in their local database with the template so that local software can be used to interrogate remote databases which have done the same.

A growing number of library catalogues are already Z39.50 compliant and use the BIB-1 template but as yet there is no widely accepted template for image material. Of course not all the relevant images are held on databases: the World Wide Web is awash with pictures and graphics. Existing search tools are rather hit and miss where text is concerned and the odds get predictably longer when it comes to finding images and other non-text material. One way to counteract this is to attach information in a standard form to each networked resource - as advocated by OCLC in March 1995 at a workshop in Dublin, Ohio. There it was proposed that a core record structure (the Dublin core) be developed and an outline structure was presented for discussion. This has been subsequently discussed and refined to the point where a second draft is now due.