DPLA Updates Metadata Application Profile to Version 5.0
Last month DPLA released an update to our Metadata Application Profile (MAP), bringing us up to version 5.0. In general, we try to release updates to the MAP infrequently so as to not have multiple versions in use at the same time. There were a couple of compelling reasons to revise MAP, and they were all primarily related to DPLA’s needs for how we store and index data. Since the last update in 2015, some changes to our core infrastructure and new technologies made it the right time to revisit the MAP and chart a new course (sorry for the pun — I promise it’s the last!).
One of the biggest areas of change in the new version of MAP is in how we store spatial data. In the last update, 4.0, we were only storing URIs to places and their “parent features” rather than explicitly storing the names of individual spatial categories like state, city, and county. While this was an elegant implementation of linked data, it relied on us having a mechanism to reconcile those URIs to get searchable, indexable data for our front end. A system like that is probably infeasible at this point for DPLA given the size of our dataset. To remedy this, we created a kind of hybrid class for spatial data that will still include URIs for place names, but will also fill in the specific indexable names for geopolitical subdivisions. The process of adding these names to records is one of the services we provide through an enrichment at data ingest.
Another big change is related to our handling of rights data. DPLA began implementing the standardized rights statements from RightsStatements.org in 2016. Our eventual goal is to have a standardized statement on every record in our collection. However, in the interim, we have had multiple rights statements in multiple classes of metadata to deal with legacy information. The revisions in the new version of MAP streamline this rights-related information into its own class: dcterms:RightsStatement. This will help us to model our own data in a more systematic way without requiring any change on the part of providers currently supplying us with records. The creation of this new class also sets us up for future developments within the international community to clarify and standardize the structure of rights metadata.
The third major area of change was to create properties to store information about IIIF (International Image Interoperability Framework) views of resources. These updates will allow us to integrate IIIF viewers directly into the DPLA website, which will in turn create a more seamless experience for users to directly access digital images from DPLA records.
A few other changes were made that were relatively minor in terms of the MAP but will allow DPLA to do big things. We defined a specific list of values for the <edm:hasType> property (see Appendix C in the MAP documentation). These are more descriptive that the basic DCMI type terms we use in the <dc:type> property, but more controlled than the freeform <dc:format> property. We also set the stage to expand and improve data about the collections that items belong to. Once these changes are implemented in records we will be able to use them to improve the user experience on our website.
Finally, we updated the Introduction to the DPLA Metadata Model to reflect the new changes, but also to add two new features: an index of all DPLA MAP documentation and an index of sample records.
This effort marked the first time that DPLA worked with a committee of Hub representatives (and one representative from Europeana) to update the MAP. Past updates were handled internally, although review was always sought by the larger community. In addition to that external review, this time we wanted to get the perspective of those who are actually trying to create and share metadata with DPLA that conforms to its model. The working group included:
- Gretchen Gueguen, DPLA (chair)
- Ivey Glendon, University of Virginia (co-chair)
- Valentine Charles, Europeana
- Robin Dean, Michigan State University
- Greta Bahneman, University of Minnesota
- Juliet Hardesty, Indiana University
- Nicole Lawrence, University of Georgia
- Kristen Merryman, University of North Carolina at Chapel Hill
- Anna Neatrour, University of Utah
- Hannah Stitzlein, University of Illinois at Urbana Champaign
- Scott Williams, DPLA
DPLA wants to extend their appreciation for the members of the working group and external reviewers (including the ALCTS/LITA Metadata Standards Committee) for their hard work and valuable contributions.
Updates to DPLA’s infrastructure to conform with MAP 5.0 will be rolled out over the coming months. The data set is currently available through an API that still conforms to the 3.1 version of the MAP. Throughout 2017, DPLA undertook efforts to update about 75% of its technology infrastructure. The coming year should see continuing updates that will bring the API up to the current standard. The changes should have minimal impact on how our partners create and share data with us. Even if a partner had adopted a previous version of MAP themselves, we should be able to easily crosswalk between the versions.
You can find the updated MAP 5.0 documentation, along with the Introduction and past versions, on DPLA’s website.