The issues associated with creating an image database pose questions not only of a technical nature for those considering setting one up, but also questions of the application of standards for cataloging, description, and selection of potential access points related to the text database component of a given image database. Imaging technology is still in its infancy. Cataloging issues related to it have yet to be fully explored. This is an area where the theoretical and practical experience of library professionals is absolutely necessary in order to allow for precise and efficient searching and easy accessibility for the user. To accomplish this, it is vital that the combined text and images that make up the database link with each other smoothly. This relates to both the imaging technology itself, as well as to the question of descriptive indexing of visual images.
Much in the literature has been devoted to issues of indexing and cataloging as they pertain to pictorial items: photographs in an archive and works of art in a museum. New applications of technology, such as automatic indexing of images raise additional questions for library professionals. The question of whether or not to apply controlled vocabulary and consistent indexing language in the textual descriptions contained in image databases would also necessitate the expertise of librarians and catalogers.
Emphasis here will be on attempting to broadly frame the issues related to cataloging of a text management system linked to an image database of artifacts that fall less into the realm of "art" and more as pieces of social or historical interest. Those questions of standardization in the development of a cataloging scheme for a particular image database are inextricably linked to defining who the potential users of a given database would be and for what purpose the database is to be compiled.
An image database of artifacts that could not be described as strictly works of art and fall more into the realm of historical artifacts, such as antiques or collectibles of one sort or another, would, like a database of images of paintings or photographs, need to have attributes in its data elements for physical description and historical context. Yet unlike descriptive catalogs or indexes for images of art, issues of iconography, use of symbols, or perceived subject matter would be less relevant in a database of "purely" historical items. That does not mean that there would be less fields needed to describe an image of an historical artifact. They just might tend to emphasize the historical nature of the item more than the visual attributes.
The criteria for the development of a satisfactory cataloging arrangement for a database consisting of images of historical artifacts will vary according to whether or not these artifacts comprise part of an archive perhaps related to a specific place, event or time period, or are perhaps part of a particular collection owned by a museum. If the historical artifacts being compiled for an image database are intended to form part of a more comprehensive catalog that lists all known artifacts which fall into a given class or category, such as all types of early twentieth-century hand-held cameras, then quite possibly the given fields in the text-based catalog for this database will need to have a higher level of detail than one for an image database which only depicts the collection of one particular archive or museum. There, the paramount concern of the cataloger might simply be records management, that is, keeping track of the items in the collection and allowing the image surrogates to be viewed instead of the items themselves, thus enhancing object preservation and meeting the display needs of the museum or archive.
The tasks of a cataloger of an image database of historical artifacts from a museum or archive will of course depend on whether the items being imaged were already cataloged as individual items, whether the entire collection was cataloged as a whole, or whether there had been little or no cataloging done prior to imaging. If the purpose of the database is to provide access to the individual items contained in a collection of historical artifacts, then detailed cataloging with access points using terms from a set of controlled vocabulary would be advantageous for more efficient access. If the purpose of the creation of the database is merely to index the items and to manage a collection for the archive or museum, then less detailed cataloging may be suitable without the use of an authority file.
If our database consists of images and text related to cameras, the data elements will likely vary depending on whether or not we are depicting a set of cameras owned by a famous early twentieth century photographer whose collection of cameras is in an archive, or whether or not it is for an exhibit at a museum that is illustrating developments in the history of the camera, or still whether the particular early twentieth-century cameras are being included in an image database intended to identify every camera ever produced.
In a situation where knowledge of the subject matters of cameras is high among the intended primary users of the database, then the level of detail in the cataloging scheme might be quite high, as it would also be for a database of art objects. But a user of a database depicting art objects might not be an art historian who needs to be able to view the image surrogate in detail or require the high level of information contained in the text record. The person doing the search might simply say, "I want a picture of a little girl holding a daisy." In that case, relatively detailed access points, both linguistic and visual could be provided to locate the image. A text-based Boolean search of "female and child and daisy" could be done, in which case the use of controlled-vocabulary descriptors in the search would aid in retrieval. As computer technology advances, a more sophisticated visually-based search relying on image shape or color might be used to locate the needed photograph.
Yet even in the case of the non-expert searcher, relatively complex access points would still be needed to narrow the search down to find the desired image. So while the level of detail in the text record might be extraneous for such a user, that level would probably need to be contained in the record. Both novice and expert would likely be using the same database at various times anyway. Therefore, it seems that any image database, used for purposes other than record management, would by its very nature require a relatively high level of complexity.
More attention in the literature on image databases has been devoted to issues of indexing and physical description for images of art objects than for images of historical artifacts. Data elements related to the physical attributes of an item depicted in a digital image, while perhaps important, would seem to matter less when dealing with items that are more likely to be seen in an historical context. However, certain historical artifacts are also noteworthy for their aesthetic characteristics, and it may also be necessary to distinguish particular artifacts from similar items - or perhaps even replicas - based on detailed cataloging descriptions of the physical attributes of that item. Therefore, a very specific classifying schema, using a broad array of data elements to describe the artifacts, based on both the physical attributes and the historical significance of the items being cataloged, would be required in creating a well-designed image database of a given set of historical artifacts.
Enser refers to the characteristics of classification outlined by Shatford Layne pertaining to cataloging issues in an image database. Therefore, broadly speaking, these classification characteristics could serve as a loose framework for the creation of a catalog scheme for a database of images of historical artifacts:
- 'biographical,' which embraces record creation, location and ownership record, value history and amendment record;
- subject, classifiable into four facets: time, space, activities/events, and objects;
- exemplified, which covers the physical form of the image;
- relationship, which covers logical linkage with other images/texts, as in preliminary sketch and final painting.1
In a potential image database of, for example, a collection of politically-oriented buttons and badges, similar questions that arose with regard to the cameras would also need to be answered. We must first determine what the nature of the collection is. Are we looking at historical artifacts collected by a well-known political figure that might encompass a part of an archive? Or are we looking at items that comprise a museum's exhibit of campaign materials from Lincoln to the present? Or still might we be looking at a collection of materials to be imaged in order to form a comprehensive catalog of all U.S. presidential campaign buttons or other political buttons? Determining the nature or context of a particular collection relates to the question of who the potential users of this database would be. And both in turn directly determine the criteria that will be used in the selection of a cataloging scheme and its various access points.
Using Shatford Layne's classifying arrangement, the category of "biographical" refers to the location of the item depicted, who owns it, the value of the item (if appropriate). Shatford Layne also includes relevant information about the creation and modification of the surrogate image ("record creation" and "amendment record") in the sphere of "biographical," which he seems to regard as perhaps not the most precise descriptive term for this class (suggested by his use of quotations around the term). "Biographical" here seems to refer relevant characteristics for a data element that are "external" to the item, which would include data on the surrogate being viewed and non-intrinsic information related to "holdings." A field for an assigned accession number in our database would also be considered part of the "biographical" category.
The "subject" class is divisible into four facets: "time, space, activities/events, and objects." As with an image database for art objects, this schema for classifying the fields of a database containing historical artifacts would be a useful way to think about them, although "objects" as used by Shatford Layne in this context seems to refer to objects depicted in a work of art such as a photograph or painting. This category would be less useful in a database of tools, such as cameras, but would be relevant in a database of historical artifacts that might have some aesthetic value to the user, such as political buttons, movie posters, or souvenir shot glasses. However, fields related to time, space (dealing with the items geographical attributes), or events would of course be relevant to historical artifacts.
The third class of "exemplified" used by Shatford Layne refers to the physical form of the image. How can we describe the object itself? What are the manufacturing techniques, materials used, the brand, size, weight, shape, condition of the object? What does it look like? This class of fields would obviously be relevant to both art images as well as historical objects.
Finally, the class of "relationship" which deals with a "logical linkage with other images/texts" would also be relevant to an image database of historical artifacts. Shatford Layne refers to a preliminary sketch and a final painting. With historical pieces, it might be early prototype and later developments, or might include references or pointers to similar images or like items. This class would be very necessary in an inventory of items in which various entries have some sequential or logical relationship to one another, a quality that would especially be relevant in an database of historical artifacts.
Obviously, Shatford Layne's broad outline of a classification scheme would have varying fields within each class depending on whether the database was cataloging images of objects in an archive, or for a more exhaustive listing of like objects. For an archive or museum, more emphasis would have to be placed on sub-dividing the fields that fall under the biographical class, as ownership and the external historical context of a particular object would be more relevant. An image database of beds slept in by George Washington would likely be categorized with less emphasis on what the bed looked like, and more on where it is, how much is it worth, who owns it, etc.
While issues of cataloging an image database of historical artifacts has similarities with those of art objects, degrees of emphasis in creating fields for a potential record would likely be different, leading to a situation where more attention would likely be given to the "biographical" aspects of the record and more detailed sub-fields needed to reflect the historical content of the material being cataloged, although art historians might take exception to a presumption that a database of art objects would have its fields related to history any less detailed than for one of a more obviously historically-based subject. While issues certainly overlap in indexing image databases of art objects and historical artifacts, more attention needs to be devoted to the specific cataloging needs of historically-based image databases in order to arrive at well-designed cataloging schemes. Since fields and sub-fields would vary considerably depending on the particular type of historical items depicted in the image database, it would seem necessary to consult with the expertise not only of cataloging professionals, but also of historians and experts in the field that is the subject of the database itself.
Bibliography Besser, Howard, "Getting the picture on image databases: the basics." Database, April/May 1995, p. 12-19.
Enser, P.G.B., "Progress in documentation: pictorial information retrieval." Journal of Documentation, v. 51 no. 2 (June 1995), p. 126-70.
Llewellyn, Richard, "Image storage and retrieval: a tool for museum collection management." Archives and Museum Informatics Technical Report, no. 14 (Fall 1991), p. 150-8.
Orbach, Barbara, "So that others may see: tools for cataloging still images." Describing Archival Materials: The Use of the MARC AMC Format, edited by Smiraglia, Richard P. New York: The Haworth Press, 1990, p. 163-91. 1 Enser, P.G.B. "Pictorial issues research: pictorial information ." Journal of Documentation, v.51 no.2 (June 1995) p.135.