Dear Ross,

Thank you for you expression of concern about searching the AMICO Library. Of course, we share it, though for a variety of reasons we are likely to take a different approach from traditional authority control in trying to resolve it.

Currently there are 23 museums contributing to AMICO. In the future there will be hundreds. Like scholars in the published literature, museums refer to artists by a variety of names. This is what led the Getty to create the Union List of Artists Names. Each of these names are equally "valid". The Getty ULAN names each have 'literary warrant'; similarly the use of a term in a museum record gives it 'museological warrant'.

In order to make links to original sources, the use of the form of name which is actually used in a source is probably the appropriate way for scholars to reference an artist (whether refering to an auction catalog or a museum catalog). Also, names are sometimes used in the vernacular but frequently (especially for well-known artists and places) given in the native language of the cataloguer. Therefore, what we need is a way for any user with one name (a - k) for a given artist, to locate works by that artist which were catalogued with a name (a - k) that refers to the same individual. This is more useful that requiring every user to know the "preferred" name, and much more practical than imposing on many different museums and scholars the requirement (sometimes inappropriate) to refer to an individual by one partricular form of their name. Similar issues arise with other terminology (as in the AAT and Thesaurus of Geographic Place Names).

We are currently engaged in an analysis of the values in indexed fields of the first 20,000 AMICO records, including comparison of these terms to the AAT, ULAN, and TGN. Next steps will involve matching to other soures such at the Grove Dictionary of Art, or Wilson index. The purpose of these tests is to identify the kinds of problems in using these vocabularies -- which range from the absence of common forms of everyday terms ("painting" or "drawing" don't appear in the AAT as such) to the absence of terms which are highly specialized or names of less well known artists. Our objective is to define a means of presenting the AMICO Library in which searchers will, ultimately, be able to use terminology with which they are familiar (ultimately it must be multi-lingual since the cataloging will be) to retrieve appropriate works.

One of the purposes of the University testbed user studies is to learn more about the intellectual perspectives of users. We hope to learn more about what kinds of access do they need and for what purposes? We hope this will enable us to work with RLG and other distributors to use thesauri and synonomy lists to collocate sources which refer to the same person, place or concept based on searches conducted using one of the appropriate forms of the term. This kind of search, through a vocabulary, should also enable users to retrieve works catalogued in a more specific way than their query (e.g., query for works in wood when some are catalogued as chestnut).

We are delighted with your suggestion that you will be available to assist us in working on this problem. Over the next few months, when the analysis of the initial AMICO Library is complete, we'll be looking for help in addressing the next questions, and look forward to being able to call on your expertise.

Sincerely,

David Bearman and Jennifer Trant