CTS URN

[gcelano]

I see that a few works that would already have a cts urn (starting with tlg) have been assigned a new one with prefix pta: I know that the equivalence is documented within the files, but I was wondering whether there is any specific reason for that.

[pharos-alexandria]

First of all, I wanted to have a straightforward system of numbering. But then, as the versions in PTA are no identical with TLG, I wanted to not get people at some point into conflict, if I have a version "-grc4" which would be different from OGL's version "-grc4", as I'm not sure whether they (or anyone else) monitors what's going on here, as we have no real central registry for all cts urns.

[mgbilby]

Perhaps some of the DH leaders in this space should formalize a registry!

The Perseus/First1KGreek team previously told me that they have tended to defer to the TLG project. But I can also see why new cts_urns are quite helpful in Patristics, particularly when so many works are misattributed (Severian of Gabala's sermons being an excellent example of this). Pseudo-Chrysostom or Pseudo-Ephrem-Graecus aren't particularly helpful either, since the compositions vary greatly in historical setting and authorial/editorial voice.

This was also an issue for our team when deciding what cts_urns to use with Marcion's scriptures. We ended up adopting the tlg author number (2958), while essentially minting new document components: 001 (Antitheses), 002 (Gospel), 003 (Apostle). The latter is a bit problematic, since it doesn't map cleanly onto the NT letters of Paul. We might eventually add an alphabetical suffix to delineate different letters within the collection, tlg2958.003a for Marcion's Galatians, etc.

The Perseus team did mention that the Scaife viewer is designed to accommodate different organizational/sponsor ids in the substring immediately after the cts_urn. The language and version information following the sponsor (eng1, grc2, etc.) indicates their project versioning, rather than global versioning.

Given this, though pta0001.pta054.pta-grc1.xml makes perfect sense, it may be easier for other DH projects to access, understand, and correlate if it were named tlg4139.tlg089.pta-grc1.xml, particularly since the tlg cts_urn author attribution in this case is correct.

[pharos-alexandria]

I simply didn't want to have a mix of file name structures which would confuse people not familiar with Perseus/First1KGreek naming scheme. If there is a tlg-structures cts urn for a text, it's always mentioned in the TEI header, so it can easily be extracted and matched. As you said, @mgbilby, TLG's attribution and division is not always helpful.

[lcerrato]

Hi,
Just an example after a test ingest.

The current Scaife Viewer display does not have a means of linking different editions without matching IDs. So side-by-side display in this example would not be possible.

Not a reason for/against the current collection organization or ID assignment but rather something we have to figure out as we move forward.

[pharos-alexandria]

Hi @lcerrato, I see the problem. Maybe for a start from my side I could also add the TLG author ID to the textgroup cts-file. For the works, the info is already in the structured metadata. Maybe Scaife Viewer could add some reconciliation using this metadata?

[lcerrato]

@pharos-alexandria
There was some discussion a while back about what to do but I don't know the precise encoding recommendations or the status of a Scaife Viewer resolution.
I think some workgroup name reconciliation may be needed in cases where there is a variant.
I'm going to create a Scaife Viewer issue to keep track of the discussion.
scaife-viewer/scaife-viewer#716

[gcelano]

Probably, the best solution is to stick to the tlg IDs, if there are no issues with author/work attribution; if there are, it makes sense to introduce a new one. The last part of the URN can be used to identify editions and also a project.

[pharos-alexandria]

• Cf. CTS URN #11 (comment) I'll add tlg-author-ids into the metadata file to ease reconciliation with the next release/update. Otherwise, I'll stick to my uniform file naming scheme.

For example, Severian of Gabala has a TLG number (tlg4139), but has not all works with identifier that are in PTA (for example the Syriac or Armenian only works); this would cause me to have a mixed scheme.