Argo report field for "created date" is missing many dates

[andrewjbtw]

Even thought this issue is visible in Argo, I think it's ultimately a DSA problem.

Describe the bug
We recently pulled apart the old Argo report "citation" field (which bundled multiple values together) and split out each of the values into their own columns.

There are many records, especially MARC-origin records, that seem like they should have a created date but instead they have none. Of over 3 million SDR records built from MARC metadata, fewer than 25,000 have a created date. Many of the items that have no created date do have a publication date, so it looks to me like we aren't mapping the "publication date" to the created date, at least when a record's origin is MARC.

The "created date" field is built off of the solr field SolrDocument::FIELD_MODS_CREATED_DATE: https://github.com/sul-dlss/argo/blob/main/app/models/report.rb#L50-L55

That field is apparently also known as originInfo_date_created_tesim, which seems to be indexed according to rules here:

dor-services-app/spec/services/indexing/indexers/descriptive_metadata_indexer_event_date_creation_spec.rb

Line 17 in 86722d4

describe 'date mappings from Cocina to Solr originInfo_date_created_tesim' do

If I'm reading that code correctly, we're deliberately not including "publication date" in originInfo_date_created_tesim. That raises two questions:

• is that correct or should we also map publication date into this field?
• is the MARC mapping correct?

In comparing MARC-origin records that have a "created date" with ones that don't, it looks like the difference can be traced back to the MARC 260 and 264 fields.

Example 1: item has a creation date

• druid: https://argo.stanford.edu/view/rf636pd0445
• searchworks: https://searchworks.stanford.edu/view/6310906

It looks like 260c is mapped to "creation date" in Cocina.

Example 2: item has a publication date and no creation date

• druid: https://argo.stanford.edu/view/mc854xw6781
• searchworks: https://searchworks.stanford.edu/view/L61302

It looks like 264c is mapped to "publication date" in Cocina.

It seems to me that for MARC-cataloged materials, the publication date should be considered a creation date, otherwise it's hard to see how these items could ever have a creation date.

User Impact
A mostly blank created date field is of limited use for items that come from MARC records, though it does seem to be functioning as expected for cocina-origin items, which are seem more likely to have "creation" dates - when they have dates at all.

This also seems to affect the Argo show page title, as the logic for generating a show page title appears to use the "citation" fields, so when there is no created date, there is no date to display.

When there is a date created, you see that on the Argo show page (screenshot from example 1 above):

When there is only a publication date, there is no date on the show page title (screenshot from example 2 above)

[arcadiafalcone]

Creation date and publication date have distinct meanings and should stay separate. The determination of date type in the MARC2MODS mapping also uses the fixed fields, so a 260 could be a date created or date published depending on what's in the leader. MARC 264 becomes a publication date only when the second indicator is 1; otherwise it maps to the event type instead of the date because it's going through MODS which does the mapping from MARC that way. If we need a "primary date associated with the resource" facet, it should pull from both the creation date and the publication date (at least) into a general date index, rather than merging one into the other. (If originInfo_date_created is only used in one place those two things may be essentially the same, but there has been confusion in the past about which dates were indexed where, so it would be good for the field name not to imply that its coverage is narrower than it is.)

[andrewjbtw]

Thanks - it does makes a lot of sense to not combine the two types of date under the same label.

I do think this field is not particularly useful for reports as long as so many records have no data. It seems like the choices are:

• create a combined field that draws from publication and creation date
• use the newly created publication_year_ssidv, which is the replacement field for the SW date facet (sw_pub_date_facet_ssidv)

I haven't yet looked into how many items have only a creation date and no publication date. Spot-checking, I keep finding a publication date is already on items that also have a creation date. An advantage of switching to the date facet field is that the reports and the Argo show page title will be consistent with the facet data.

We aren't putting originInfo_date_created in a facet, as far as I know. It only seems to be in the report field and the Argo show page title.

[arcadiafalcone]

Photograph collections use the creation date instead of publication, as do manuscripts.

[andrewjbtw]

I see - that actually looks like it's been taken to account in the "date"/SW date facet, despite the solr field name using the word "publication".

This search for a photo that has only the creation date does show a date in the facet.

That makes me think that:

• the report should use that solr field
• we may want to rename that field more generically