BCAUL Pilot project: Crosswalks
Crosswalks establish relationships of equivalence between data elements in two or more standards.
Practical purpose: required when an application built around one standard needs to interact with data structured according to other standards.
- Import: take in data originally structured by a different standard.
- Export: output data in form structured by different standard, typically for use by application built on that standard.
- Design: expand user base of application by adapting it to other descriptive communities (use different standards to describe same or different types of materials).
Import and export involve devising rules for handling data and implementing these as methods or scripts in the application. Design addresses both database architecture (more tables or fields needed?) and user interface (layouts / forms for data entry and presentation).
The BCAUL project requires crosswalks for each of the above purposes (import, export, design).
- ICA-AtoM is built around ICA descriptive standards.
- The application needs to be able to import and export data in a number of exchange standards: EAD, MARC21, MODS.
- The software must be adapted to Canadian descriptive standards (RAD) for use by Canadian archivists.
While outside of the scope of the BCAUL project, it should be noted that the underlying software on which ICA-AtoM is based (Qubit) is intended as a metadata toolkit for describing collections of information objects -- archival materials, but also museum artefacts, works of arts, exhibition supporting material, and publications. These other, non-archival types of material may require other descriptive standards and crosswalk work has included Dublin Core (DC) and the Categories for Descriptions of Works of Art (CDWA).
The relationships between data elements identified by crosswalks fall into three broad types:
1. Full identity
- 1 to 1 relationship between an element in the source standard and the target standard.
- E.g. RAD Custodial history = ISAD(G) Archival history = EAD <custodhist>.
- Conversion is "lossless": no information is lost when the element is converted from the source standard to the target standard; and when it is converted back from the target to the source (sometimes referred to as "round-tripability").
2. Partial identity
- Partial overlap between the elements in the source and target standards.
- May be a many to 1 relationship (data segregated into many elements in the source standard are brought together in a single element in the target).
- May be a 1 to many relationship (date contained in a single element in the source standard is dispersed into several elements in the target).
3. No identity
- Elements exist in the source or target standard that have no analog in the other.
- May be 1 to 0 (source standard element that has no equivalent in the target).
- May be 0 to 1 (target standard element that has no equivalent in the source).