In her research about a virtual museum, Kenderdine (1996) pointed out that the design of an online collection should seek to embody the unique characteristics of the discipline while fundamentally transforming the way the data is accessed, presented, updated, debated, and researched. The design of an online collection should begin with an investigation of the unique characteristics of the discipline and the potential use of the collection.
A digitized fashion collection on the Web creates a completely different dynamic for a material culture collection, overcomes conservation problems, and makes these artifacts available to large (including new) audiences. Museum registrars, curators, gallery directors, art historians, art conservators, art educators and students are likely to be the primary users of a digitized historical fashion collection. Other users may include collectors, insurers, art administrators, students in other disciplines, critics, artists, wardrobe designers for movies and plays, art librarians, editors, publishers, restorers, suppliers, writers, and the public.
A major benefit of digitizing a collection is to make it more accessible to a wide range of users. Digitizing is not merely an appropriate and desirable alternative for museum exhibitions. The ability to search and browse the digitized collections through various access points will greatly enhance knowledge and use of the collections. To a large extent, many museum websites have achieved the goal of serving as an alternative for museum exhibitions. Images of costumes are usually displayed or linked with curators' prefaces or notes. In most cases, it is the explanatory material that ties the diverse pieces together. On other occasions, materials are arranged chronologically or geographically, or are grouped under broad categories, while images are accompanied by various kinds of surrogates. Few of these websites, however, provide searching options in addition to browsing options; therefore, the power of searching through different access points according to various needs has not been efficiently realized. The weaknesses of lacking search functions and support for evaluation and analysis have to be overcome.
An examination of fashion collection websites suggests some of the reasons for such weaknesses and limitations, which are largely related to the role of surrogates in resource discovery. Surrogates are cataloging/indexing records that describe the actual resources and inform the user of how to access the objects themselves. Surrogates may be richly detailed in their identification of significant object attributes and relationships, or be so brief that their primary function is to indicate the existence and location of an object only (Younger, 1997). Among the problems observed were the following:
1. Most of the websites lack structured surrogate records; hence, there is no foundation for building a searchable collection;
2. Websites which use a structured format to form textual surrogates do not seem to have followed a format consistently, while the types of information included by the surrogates vary from item to item and from collection to collection;
3. Content presented in the surrogates is often only an online version of exhibition labels or captions of costume objects; and
4. Information contained in the surrogates is not fully searchable. It seems that the surrogates and search mechanisms within the same websites were not designed or implemented coordinately. For example, the Museum of Costume of Bath, England, (http://www.museumofcostume.co.uk/) has included very useful elements in their records such as color, sex, technique, material, pattern, and provenance, but none of these are made searchable as of January 15, 1999.
Realizing the common weaknesses of current digitized fashion collections and some of the causes, the author delved into an analysis of related projects and research to identify the elements to be included in a catalog record. It is always a formidable challenge to determine how to verbalize the contents of three-dimensional objects, how to choose particular aspects of these objects to index, and how to interpret the subject matter that these objects represent. Hastings conducted a study in 1995 on research queries posed by art historians to an art image database. In order of frequency, the major classes of queries were Identification, Subject, Text, Style, Artist, Category, Comparison, and Color. Jorgensen (1996) used previously identified attributes to create a template for image description. The attributes were identified as perceptual (responding to a visual stimulus, such as color or object), interpretive (requiring knowledge or inference, such as style or atmosphere) and reactive (responding on a personal level, for example linking the image). In her discussion of image retrieval, Layne (1994) indicated needs for empirical research into attributes, other than subject and the traditional author and title, that would be useful as access points. She divided attributes of images into four categories: Biographical, Subject, Exemplified, and Relationship. Lunin (1994) listed 24 fields that are relevant for a database describing fiber art. Special fields included technique, structure(s), theme, style, period, color(s), dyes used, texture, decoration, surface embellishment, design, provenance, and exhibition history. As Taylor (1999) summarized, a single museum record "has many more fields than does the usual library catalog record. Some fields that might be needed for art objects that are not used in libraries are: Material content, Technique(s), Studio of origin, Type of equipment used, Color(s), Texture, Design symbolism, Provenance, Exhibition history, Installation considerations, Appraised value, etc." (p.10).
Many fashion museums' websites keep interesting biographies for their collection objects. These biographies are often important because from the "birth" of an object to its "travel" between different owners and with different exhibitions, often notable characters and stories arise. However, outlining the history of an item is not enough to meet the needs of primary users of this digitized historical fashion collection. Denoting a fashion item's style, fabric, pattern, technique, and measurement will provide a foundation for identification and comparison between fashion items and also for inducing and deducing the generalizations, principles, and consequences that relate to fashion objects. Information about the function and construction of a dress will speak to the impact of fashion on behavior. With additional indicators of temporal and geographical significance, these characteristics also will further our understanding about a particular culture and/or society and identifying cultural influences in the fashion history. To satisfy researchers' interests of abstraction and classification, it is necessary to provide history, cultures, and additional background information. To support interpretation at this level requires presenting meaningful terms, known as "subject terms" or "subjects", which indicate the social and cultural context in which an individual object was produced, as well as narrative descriptions. Subject and topical terms, proper names, temporal and geographical information are indispensable in research and learning. Also, these elements amount to natural links to tie together the diverse items in this or other collections. Along the lines of AACR2 and USMARC format, these necessary elements for descriptive and interpretive can be categorized into descriptive information and subject information. Given that a museum's accession record usually contains an object's biography, the author intentionally set up three categories:
1) Registration Information: includes accession information, nature of the item, patent information, credit, source/donor, mode in which the item was acquired, primary designers and manufacturer, owner(s), dates, exhibition history, etc.;
2) Descriptive Information: includes information about a costume's type, style or particular influence, accessories and parts, dimension, dye(s), fabric, pattern, surfacing and technique; and
3) Subject/Topic Information: includes subject terms that indicate the function of a costume, its primary cultural influence, its period of fashion design, and notable terms beyond a general description. In addition, KSU Museum curators suggested that records provide interactive links to other sources of information. For example, hyperlinks to other related records about parts, accompanying pieces and accessories, and hyperlinks or pointers to citations or other forms of references related to the object being described.
The set of resulting elements is shown through a sample record in Appendix C. Some of the elements may be necessary only to meet higher level users' expectations. Whether the considerations above could be applied to other fashion collections, or to other types of objects, requires future testing.