DC Tabular Application Profiles

Many communities today create application profiles (APs) for their data. Application profiles provide the rules that govern the creation and reuse of metadata instances. Their function is both to explain the metadata but also to potentially constrain the metadata so that correct usage can be determined. A single profile can serve a variety of needs: metadata creation support, metadata validation, metadata exchange, metadata selection, and mapping between metadata from different sources. Application profiles need to be shareable so that data exchange between communities of practice can take place. There is, however, no current standard for the creation of application profiles such that they could be understood outside of the community within which they have been developed.

Dublin Core Metadata Initiative has had profiles on its agenda since 2000, including a framework and a constraint language. This year a working group was formed to develop a simple, core vocabulary for the information carried in application profiles. Because profile documentation often takes the form of tables or spreadsheets, it was decided to create the initial vocabulary specifically in support of a tabular model: a table or spreadsheet with an underlying format of comma separated values (CSV). However, the data elements can support other data formats.

In creating this first version of the application profile we assumed that the profile would be describing and constraining instance data that is formatted as RDF. For purely practical purposes we felt that this would be the most useful for the community served by DCMI. A subsequent step could include testing this approach for non-RDF data if there is an interest in that.

The proposed vocabulary consists of only 12 elements yet the model supports entity descriptions, property listing, cardinality rules and value rules. There is a draft primer for the proposed profile template and a vocabulary definition document. There are some proofs of concept in the "examples" folder.

Discussion and work plans are at the project mailing list and in Github issues. We welcome any and all participation in the work.

To gain community input, an open meeting was held on December 17 on Zoom and the video may be viewed on YouTube.