Use case 1: Metadata crosswalk

This is a short extract from ONIX/MARC21 mapping by Bob Pearson of OCLC, available from ONIX mappings to MARC on the EDItEUR website. It is an existing published crosswalk between two schemes, which is used by others in creating transformations from ONIX to MARC records for library use:


This extract contains two ONIX vocabularies (lists on the left hand side) with the corresponding MARC terms for use in specific records (on the right). Some "special cases" are indicated below the lists. In the first case the mapping is from one vocabulary to another, with codes on both sides in this example; in the second case the mapping results in the creation of a text description in MARC.

This use case raises two fundamental problems:

  1. What happens when a change or addition is made to either vocabulary on either side?
  2. On whose authority is the mapping made?

The answer to question 1 (change) is that someone has to manually update the crosswalk whenever a new change is published in one of the standards. This is already a significant issue, and will become a major problem with proliferation of schemes and crosswalks. We have seen that ONIX vocabularies are growing at the rate of up to 10 percent a year, and we expect MARC to publish some significant changes in 2009. There are no guarantees that any published crosswalks will be maintained, and there are no formal mechanisms in place to do so.

The output of this project will enable a creator of a crosswalk guide, an XSLT or similar transformation, to embed links to the RDA/ONIX Framework which will enable the crosswalk to call on (or, if preferred, to generate) the latest term-to-term mappings from the RDA/ONIX namespace at the time that it is accessed, instead of having to manually encode the mappings and maintain them thereafter.

The answer to question 2 (authority) is unknown. The status of such crosswalks is almost always informal. As it is published on the EDItEUR site, we may presume it has the support of the ONIX publishers; its status in relation to MARC is not clear.

This is the case with most existing crosswalks: mappings are entirely a matter of individual judgment, and require re-analysis of the schemes and vocabularies on each occasion. Whereas a standard has governance, mappings almost universally are proprietary or "open source", so for the user it is simply a matter of "which do you trust" or even just "what is available". A standard may be carefully agreed and published, but its editors or governors rarely have any authority over how it is represented through mapping in a transformation.

The RDA/ONIX Framework proposal provides the opportunity for the governing authorities of a standard to participate in establishing an authorised mapping of its vocabularies into the ontology, and therefore into other mapped vocabularies.

Participation of authorities in the mapping process also improves the quality of the results. The success of this has already been demonstrated in a limited way in the initial creation of the RDA/ONIX Framework involving experts who were editors or representatives of a number of the participating standards. The process itself can also help authorities to identify and amend weaknesses or gaps in their standards.

These two issues – change management and authority – are key quality issues for metadata, and this project will put in place the means of achieving them, which can apply to existing crosswalks and to new ones. Both RDA and MARC are expected to publish substantial sets of vocabularies in 2009, and there will inevitably be a flurry of mapping and crosswalk activity in relation to one another and to some of the other standards in the scope of this project. It will be very beneficial to have the RDA/ONIX vocabulary in place to meet that challenge to support and enable the next generation of crosswalks.

This use case also identifies examples of common mapping issues which can be managed in the RDA/ONIX Framework mapping process:

  • mapping to literal values rather than vocabularies (ONIX tag name to MARC description)
  • qualified mappings ("if second character i"…, "if any of the following are present …")