Site Visit to the University of Michigan Museum of Art

Cory Bandt, Yu-Ling Fan, Maia Jin, Cara List, Vlad Wielbut

ILS 603, Image Databases

Professor Howard Besser

Background Information

The Card Catalog

Prior to computerization, all intellectual control of the Museum's holdings was accomplished through the use of a paper based filing system composed of three series of records. The principle series is a master file containing descriptive information and a small black and white photo of the item, the second is a vertical file with additional non-standard information, and the third is a locations file. All files within this system are physically arranged by accession number. Intellectual and physical access to information in the filing system is provided solely by accession number.

Catalog cards in the master file include fields for accession number, general category, nationality, attribution, title, collection level information, date, description, and donor information. The back of each catalog card contains additional acquisitions and collection information if any. Controlled vocabulary was not systematically used for description in the master file with the exception of the general category field. This field recorded medium, technique, or type of object being described. The following twenty-seven possible entries in the general category field are identified:

The vertical file contains copies of transfer documents, other photographs of the item which may have been taken, citations to relevant published materials or copies of unpublished research on the artwork, and more detailed collection level information. The vertical file serves as a repository for all "other" information which will not fit neatly on a 3'x 5' index card. The nature and quantity of information in the vertical file varies greatly across holdings. Lack of standardization in the vertical file renders automation of this information problematic.

The locations file is composed of index cards with accession number and current locations information for each item in the museum's holdings. This file was updated frequently to reflect loans and rotations of materials.

The Argus System

In 1990, the Museum realized that they needed a computerized catalog for easier access to the records. They bought the Argus system from Qustar Systems, and chose this system primarily because it promised to allow donor information to be linked to membership information and development files. They hired four people to enter data on about 12,000 objects and drew this information directly from the card catalog. None of the bibliographic information was entered at this time, however, because of the enormity of that task and the time and money required to complete it. On only the most famous and frequently researched items was bibliographic information entered. On the other works, the notes field may include a reference to the bibliographic vertical files.

The fields that the Argus system includes are:

The subject field was a new field, and caused the most difficulty in cataloging. Ultimately, the data entry people simply guessed what headings were appropriate from what could be determined from the card catalog records. There is very little standardization, even though this field is supposed to contain words chosen from a 90,000 word lexicon which is based on subject headings from the catalogs of past Argus users. Another downfall of Argus, is that the search capability is very limited. Searches are based on phonetic pronounciations and frequently words that sound alike are brought up incorrectly in searches, and this was very frustrating for the catalogers. The Museum currently uses the Getty Art and Architecture Thesaurus for cataloging, but this is a relatively new development.

Medium became the only standardized field, as it was based on the Medium choices from the card catalog records. However, a larger variety of Medium choices were entered, and this means that the Argus records now differ from the card catalog records. (Note the Medium field in Figure One and Two). However, through the transition to Argus, the accession number field remained constant, and this is still a unique number by which to retrieve records.

Another shortcoming is that the Argus system is not very user friendly, and the donor records were not linking to the membership information to the Museum's satisfaction. Finally, one of the biggest problems with the Argus system was that no images were attached to the data records.

The embARK Catalog

In July of 1994, the Museum realized that they needed to expand their system. They owned the core module of Argus, and there was a module available to include images however that would be an additional purchase. From a more technical stand point the imaging module was rumored to have glitches, and the compression level was too high to do any thing but slow the system down to a virtual standstill. It had been their experience that even making minor changes in the system caused major upheaval. What would happen when they made such an enormous change as adding images? The problem was solved when Bill Hennessey, the director of the museum, was impressed by a demonstration of an early model of the

Good news is that the path has already been somewhat traversed by the Department of Education's Image Browser Project, to which the Museum of Art was a major contributor; many of the issues involved with putting large image database on-line (image compression, search engine, browser interface, memory requirements, costs, etc.) have had to be resolved and significant experience has been gained in the process, and although the audience and, therefore, the needs of the Museum are significantly different, some of this experience is clearly of value to any image database developer.

Lastly, QuickTime VR capability of Embark was mentioned briefly during our first discussion with Museum's representatives: given the fact that many artifacts in their collection are clearly intended to be viewed as 3-dimensional objects, wouldn't it be nice to provide this "look around" viewing of artwork on-line? Sure it would, although some anxiety exists as to the whether it would not discourage people to visit the actual Museum. However, costs of implementing this new and unfamiliar technology make it clearly a "wave of the future".