OAI metadata mapping for journal articles (for harvesting by Primo)

[dolsysmith]

Several GW-published or edited journals are archived as collections in GWSS. To enhance discovery of these collections, we can harvest the metadata for indexing in Primo.

To accomplish this, I think we'd need to do the following:

• Create a new OAI mapping (similar to the oai_dc_etd mapping for ETD's) to include Dublin Core fields at the article level indicating the journal title and issue information.
• It may facilitate the creation of these fields to create Journal Issue works as the immediate parent of all individual article works.
• Create a new OAI set that contains just works nested under Journal Issue works. (Dependent on the previous.)
• Create an XML mapping in Primo to translate this DC metadata to a Primo record with the appropriate fields.
• Create a new Primo import profile to target this set of works.

Decision points:

1. How do we want to handle cases where the entire issue is treated as a single work? Do we want records for journal issues in Primo?

[dolsysmith]

Notes from Schol Comm's discussion with Matthew (RDG).

General approach

• Ingest GWSS works into Primo for individual articles published under GW auspices
• For journals where we have GWSS works only at the volume/issue level, RDG can create MARC records in Alma and link to the GWSS collection.

Some preliminary thoughts on implementation:

• The best place to capture the title of the journal and the volume/issue for articles is in a GWSS work of the GwJournalIssue type.
• Individual articles would be assigned as child works of the journal-issue parent.
• To create the OAI set, we can take one of at least three approaches:
  1. Find all works where has_model_ssim is GwJournalIssue, and for each journal-issue result, retrieve the works listed in the member_ids_ssim field.
  2. Create a faceted field that indicates whether a work is GW-published (or something analogous). Faceting on gw_published:true will yield the relevant works.
  3. Create a new administrative set just for GW-published articles and populate it with these works, and have the ListSets behavior generate sets based on the admin set. (I think this would be possible.)
• For reasons outlined below, I believe that the 2nd or 3rd option may be technically the less complicated to implement than the 1st.
• To create the Dublin Core metadata for each work, we can get the journal title and volume/issue metadata from the parent_works attribute of the Fedora GwWork object. (Note that the Solr result for a work does not contain a link to the parents, only the children.)

Other issues

• To expedite assigning works already in the repository to journal issues, it would be expedient to use Bulkrax. Workflow:
  1. Create Bulkrax Exporter for all the works in a given journal collection.
  2. Create the Journal Issues in the UI for each volume/issue.
  3. Assign the appropriate issue to the parents column of the exported CSV.
  4. Re-import the CSV to update the metadata.

[dolsysmith]

Technical notes

• Need to create an OAI set that contains just records for journal articles for harvesting in Primo.
  • One option: customize behavior of the GwJournalIssue set, so that it displays child works and not the parent work.
  • Alternatively, the ListSets behavior of the Blacklight OAI provider gem can be customized. But this behavior leverages faceted fields to produce sets, so without overhauling the logic entirely, we'd need a facet that corresponds to that set of journal articles, which we currently lack.
• To test:
  • Observe to_sets behavior on an instance of a SolrDocument.
  • I'm guessing the find method in the SolrDocumentWrapper class is the pivotal method. (See also here). It may be necessary to override this behavior, or one of its dependencies.
  • The set query assumes the form of a simple Solr filter on a facet field.
  • Because the find method ultimately leverages the search_builder parameter chain, we'd probably have to deconstruct this to create the more complex search required, i.e., to target records of type GwWork that are members of GwJournalIssue records.
  • Can we create more than one type of set by providing multiple facet fields? Yes, maybe?

[dolsysmith]

Update: after conversation with Schol Comm, recommendation is to use a separate admin set to identify OA-published articles for harvesting by Primo.