Just as standardization of descriptive vocabulary is important for the long-term success of digital museum projects (see Besser, 1999), standardization of metadata is important for the integration of those projects into the web. For users to have access to the full potential of the web, users' computers will need common languages for communicating with each other and with the countless servers, databases, digital libraries, and repositories out there. Many information scientists are working on a "semantic web" that will allow this communication (see for instance W3C, 2002, Mar. 7), but for now it is mostly a dream.

There is a confusing array of data standards and metadata schemas available to museums:

I believe that simplicity will be the key, at least in the beginnings of the semantic web. Most museum staff are not information professionals; many are volunteers. Many museums can devote only tiny portions of their budgets to collection documentation. Until there is an international, cheap, out-of-the-box software system available (like SPECTRUM), the museum world will be better off defining an easy, minimum standard.

That said, I will concentrate on selecting three things for the Drexel Digital Museum: an interoperability standard, core data categories, and a metadata standard.

Interoperability increases the value of the data in individual repositories by allowing easy interlinking of databases. The two major choices are OAI Protocol and RDF.

OAI Protocol is an XML schema providing additional structure to Dublin Core metadata. Although it is being developed for use by archives, CIMI has tested it for museum use and found it simple to implement (Perkins, 2001).

RDF is an XML schema also, but not necessarily dependent on Dublin Core. Furthermore, it is being developed for use by all providers of web-based data and is thus a more universal standard.

Both OAI Protocol and RDF are an "envelope" around metadata, routing queries to the correct namespace, which can translate metadata inside the "envelope" into whatever form is readable by the querying machine. Both are simpler to implement than Z39.50 and are meant to supersede it, though neither has Z39.50's proven track record (Miller, 1999). Because RDF is a more universal standard than OAI Protocol, I would recommend it for the Drexel Digital Museum. However, since the museum director has already invested considerable time in OAI (see Martin, 2002), the museum should follow its protocol. As OAI becomes more common, it will probably become interoperable with RDF.

Various museum groups and other organizations have set forth core categories of data which must be compiled to describe or represent an object. These usually include creator, object type, a description, and several other categories. The data categories an individual museum chooses will depend on its goals and affiliations.

For the Drexel Digital Museum I recommend the following:

• CIDOC Information Categories, because they are an international standard
• AMICO Data Specification, because it is becoming a U.S. standard and because it is a prerequisite for participation in the AMICO consortium, an important educational resource
• Object ID, because it is recommended by national and international law enforcement agencies

VRA Core 3.0 and CDWA are both important standards, which are, nevertheless, almost identical to CIDOC and AMICO. The REACH Element Set could become important if Drexel University joins RLG.

Dublin Core has become a universal metadata standard. OAI Protocol depends on it and RDF assumes that most data providers will use it.

AMICO Data Specification has its own metadata standards for each data category, though the standards are not terribly different from Dublin Core.

PRESERV Metadata store information often ignored or poorly documented in digital projects, such as digitization process and image viewing requirements.

USMARC is a huge and complex standard and is beyond the scope of this paper. Nevertheless, USMARC fields can generally be mapped onto metadata elements of other standards if necessary (see Baca & others, 2001).

For Drexel Digital Museum, I recommend the following:

• Dublin Core, following CIMI Dublin Core best practices as appropriate
• AMICO Data Specification, as a second metadata set
• PRESERV Metadata, for metadata representing technical details of imaging

implementation

Currently, items in the Drexel Digital Museum do not have extensive metadata visible to the public (see html code for web display of garment number 3607 a,b, below):

However, Olson & Javins (n.d.) have developed a cataloging template for the Drexel Digital Museum based on the Dublin Core. I assume that this template has been used for cataloging the collection even though the results are not visible at this time.

Marcia Lei Zeng is doing interesting work on museum object metadata, specifically for objects in the Kent State University Historical Fashion Collections. She has designed a data entry interface for cataloging objects (Zeng, 2001, especially http://circ.slis.kent.edu/mze ng/vra3template.htm). Once the data has been entered, the cataloger can choose to have it processed as VRA Core 3.0, USMARC, or Dublin Core. Zeng's documentation is not entirely clear, but it appears that the cataloger may also choose to have the data elements stored without processing according to an outside standard.

This last option seems most sensible to me. Potentially, almost any metadata standard could then be supported by programming a "crosswalk" that would take pieces of data and place them in the proper metadata "slots." For instance, a crosswalk for the title of an object (following Baca & others, 2001):

I have created a rough metadata crosswalk (available as a separate file, crosswalk.xls) for data elements required by the various standards I have recommended.

In my opinion, the specific standards adopted are less important than the act of adopting a standard. As long as a museum has implemented at least one generally recognized standard, it should be able to share data with other museums.

next page: data entry interface requirements

back to project home page

Donny Smith
May 2002
[email protected] or [email protected]