As discussed in previous sections, it was anticipated that more than one large data repository would be interested in KSU's digitized historical fashion collection, but each repository would require a different metadata format. As a result, mapping tests were conducted and two mapping tables were prepared for conversion from the extended VRA Core records to USMARC and Dublin Core records. At the beginning, it was thought that mapping from a more detailed format to a simpler format would be easy. However this assumption was not correct. The VRA Core specifications prepared by the Visual Resources Association Data Standards Committee suggested mapping fields of USMARC as well, though sometimes they were not appropriate when applied to a fashion collection. A useful reference was Caplan and Guenther's (1996) work which noted problems and questions on mapping Dublin Core data elements to USMARC format.

The mapping test among the elements of the three formats encountered difficulties such as one-to-many, many-to-one, no close match, and partial match. A software program for automatic mapping may handle some elements like the VRA element SUBJECT and NOTES. For example, various types of subject elements in the extended VRA format can be mapped against various MARC 6XX fields, or they can all be put into DC's SUBJECT element. But this circumstance does not ensure that the VRA SUBJECT elements and MARC 6XX fields or DC.SUBJECT are equivalent. VRA has STYLE/ PERIOD/GROUP/MOVEMENTS, and NATIONALITY/CULTURE as separate elements in addition to the SUBJECT elements. The same situation can be found for the NOTES elements. Various VRA NOTES match particular MARC 5XX fields; or all of them have to be put into DC.DESCRIPTION element. However a number of VRA elements other than NOTES also need to be mapped into MARC 5XX fields. Some elements are mapped to 5XX fields or DC.DESCRIPTION simply because there is nowhere else to put them. In a number of other occasions, one VRA element will need to be mapped into two MARC fields or two MARC subfields.

In VRA Core several elements contained uncontrolled keywords (such as the word "American" in Nationality/Culture element). When such a term is to be converted to a MARC 650 $z subfield or a 651 field, the conversion from an uncontrolled term to a LCSH heading (such as from "American" to "United States") will need to be considered. There is also a need for checking related fields (such as a "043 $a n-us---" for the above example.) Punctuation, capitalization, and order of terms may also require changes when mapping terms and phrase from one VRA element to more than one subfield. A fuller discussion of mapping issues in a later paper will be based on continuing and larger scale testing. It is clear that a great deal of the conversion process still requires human judgment. Both MARC and DC have fields that do not map easily to the VRA Core set. In general, the high-level of specificity for USMARC and the high-level of generality for Dublin Core have created various mapping difficulties. Since the VRA Core allows description of both the original work and reproductions of the work in one record, Dublin Core's orientation to document-like objects has made it more difficult to accommodate all the VRA elements.