Finalize data model for encoding related subjects

[wlpotter]

@davidamichelson I am using 133 as my example as it had examples of all the categories of related subjects: https://caesarea-maritima.org/testimonia/133

I have elected to use a tei:listRef for this draft/demo. I don't think this use fits perfectly with the TEI's intention for this element, but I couldn't find anything better. tei:textClass could be ideal for the themes, but for the others (geography, prosopography, and related-texts) it may be too unwieldy to define and maintain a taxonomy.

Here is the XML snippet I created (/TEI/text/body/listRef, following the tei:bibl elements):

      <listRef type="related-texts">
         <ref target="#">John of Nikiu 66.2</ref>
      </listRef>
      <listRef type="geography">
         <ref target="https://pleiades.stoa.org/places/658381">Antioch on the Orontes</ref>
         <ref target="https://pleiades.stoa.org/places/1001940">Palestine</ref>
         <ref target="https://pleiades.stoa.org/places/629035">Caesarea Mazaca</ref>
         <ref target="https://pleiades.stoa.org/places/628949">Cappadocia</ref>
      </listRef>
      <listRef type="prosopography">
         <ref target="#">Augustus, emperor </ref>
         <ref target="#">Herod I, king</ref>
         <ref target="#">Archelaos</ref>
      </listRef>
      <listRef type="themes">
         <ref target="#">Hellenistic History</ref>
         <ref target="#">Roman History</ref>
         <ref target="#">Geography</ref>
         <ref target="#">Government and Law</ref>
      </listRef>

A few notes

• I used Pleiades to fill in the targets in the geography section to use as a test, though per Joe's email we will have to figure out what to use for prosopography (is VIAF only for authors?) and for related-texts. For themes we will likely need to define our own taxonomy.
• The # in the @ref attribute is merely a placeholder for validation purposes; we can leave those attributes empty if we want
• From a data transformation standpoint it should be pretty easy to convert the notes to listRefs
• For prosopography we talked about having our own place name versus getting it from Pleiades. I can't remember how we left that, but it might get a bit complicated (though tei:ref elements are relatively permissive with what is allowed in them: https://www.tei-c.org/release/doc/tei-p5-doc/en/html/ref-ref.html)
• We will also need to deal with distinguishing between related-texts references to other testimonia and those not in our db.

[wlpotter]

Here's a note from Joe about the prosopography links:

Quick question to ponder: when/if we install a "related subjects" content area in each record that leads users toward the various people mentioned in the testimonium ("prosopography"), I dont know of an online source for which we can insert URIs that parallels Pleiades for the places ("geography"). There might be something out there I dont know, but I dont think there is a PIR or PLRE on-line; I have seen the PBE out of King's College which is very nice. I hesitate to make links to Wikipedia or Encyclop Brit. Something to ponder?

[wlpotter]

I will make a separate issue (#108) for cleaning up these tei:note elements before we decide how to finalize their encoding.

[wlpotter]

After #103 is fixed, I can dump a list of the in-use related-subjects terms for the editors to look at, revise as needed, and turn into a controlled vocabulary. This can be used in other records that have not had their subjects encoded; and it can be used to create taxonomy URIs for browsing, etc. using, perhaps, tei:relation elements.

By the way, we could also cross-walk with our taxonomy keyword URIs if we wanted to.

[davidamichelson]

Data has been updated by #108

[wlpotter]

To implement this:

• Make a spreadsheet of existing related subjects. Have editors/students review to reconcile duplicates and then update data with controlled and normalized terms
• Then match controlled terms with:
  • related-texts use our own testimonia
  • geography use Pleiades
  • prosopography use wikidata
  • themes us LCSH
• update existing data with the matched URIs
• create an app issue for the facet functionality
• update the input form to have a dropdown menu to select related subject terms
  • the post-processing script will add the URI
  • we will have the dropdown allow user suggestions, which we will then need to catch and output a warning that those weren't found in the URI lookup

[wlpotter]

FYI spreadsheet is here: https://docs.google.com/spreadsheets/d/1pdVVVt11X-lmiRAs6D6e9Yiis8ZqN11JrHvPFcO_CW8/edit?usp=sharing

The following will be needed:

• review the 'distinct' tabs for each related subject type to catch variants of the same item (place names, especially). This will let me run a batch update that normalizes the controlled vocabulary term (as long as we don't delete the 'CurrentName' columns)
• add the URIs to those distinct tabs
  • One trick here will be the cases in the 'related texts' that are either not in the Testimonia database or are for the work-level rather than testimonia. I think we may want to rethink our sources? Could just use the Perseus Catalog but not sure of coverage

[wlpotter]

@davidamichelson do we want to bring this up again in January? The data is currently still just in //body/note/p elements and could use some normalization

[davidamichelson]

@wlpotter Yes, let's implement this change to list/item, great suggestion.

[wlpotter]

Note to self: update this to include a desc element in each listRef that provides a human-readable name of the category for displaying it on the frontend. E.g. //listRef[@type="related-texts"] would have <desc>Related Texts</desc>

[wlpotter]

Note: record 381 has been updated by hand as an example for display; I've retained the old model of encoding as a comment, so revert back to this before batch updating to catch any normalization of the subject terms

[wlpotter]

Script for ingest is: https://github.com/wlpotter/sandbox/blob/main/c-m-scripts/update-testimonia-related-subjects/ingest-caesarea-testimonia-related-subjects.xq

Waiting on final word that #183 is complete