CMR: the Canadiana Metadata Repository schema

Canadiana has published its internal metadata specification: CMR (Canadiana Metadata Repository). All metadata contributed to the Canadiana Discovery Portal is converted into CMR. This process involves mapping descriptive and full-text fields to a common set. It also involves normalizing certain fields, such as publication date, media type and language, to a standard set of types in order to allow for faceted browsing based on these criteria. CMR also maps parent-child relationships, allowing a complex item or set of items to be broken into constituent parts . For example, a book can be indexed as a set of records: one for the book and one for each page, allowing for searching and access at the document level and at the page level. CMR maintains the parent-child relationship between the document record and child page records, as well as the sequential ordering among the page records. The same holds true for series and issues.

Although originally developed for internal use, we published it in the hopes that it might be useful to others as well. Metadata for the Discovery Portal can be contributed directly in CMR format, though we continue to accept whatever other formats you might have (Dublin Core, MARCXML, and so forth) and do the conversion ourselves. CMR is moderately complex: it has more structure and greater constraints than Dublin Core, but is still a fairly lightweight and easy-to-understand, purpose-specific format.

CMR is an intermediate data format: records are stored in CMR and transformed into a Solr schema for indexing into the Discovery Portal. This two-stage process allows us to make frequent changes to the Solr schema without having to re-map every contributor’s metadata, with its unique quirks, to the new schema. Instead, we can take the already-normalized CMR records and generate updated Solr records using a single, straightforward XSLT stylesheet. It also means that all contributed metadata is available in a structured, application-neutral format rather than a messier application-optimized format.

If you’re planning to contribute your metadata to the Discovery Portal, or if you are in the process of working on a metadata schema for your own digital content, I encourage you to have a look at CMR. Naturally, we also encourage feedback, comments, and suggestions for improvements in future versions.