tlg5026.tlg001 tlg9010.tlg001

[gcelano]

These texts are larger than 10MB and give a hard time when post-processed. It may be worth considering splitting them.

[mgbilby]

I'm experiencing something similar working on the OGA derivative of tlg5026.tlg001. My post-processed conllu file exceeds 100 MB, so it won't sync with Github through standard means.

It appears that this work covers Iliad books 1-6. Splitting at the start of book 4 might be sensible. The second part would thus begin at token 457059.

Also noticed that tlg5026.tlg001 has unconventional single right square brackets to offset quotations from the Iliad prior to the commentary. Can the text be adjusted to wrap these quotations in quotation tags of some sort?

The text also has lots of parenthetical citations (pages, fragment numbers) that may also be considered as candidates for wrapped in tags so as to avoid syntactical parsing errors downstream.

[lcerrato]

@mgbilby @gcelano
This is an interesting problem.
We wouldn't generally split a work arbitrarily as this has cataloguing, presentation, and URN implications. (In fact, we've been merging previously split files). I can ask @AlisonBabeu to raise this with the OGL group at the next meeting (I'm not able to attend regularly). I think we want to avoid arbitrary divisions in favor of those suggested by print, so 4 books would suggest 4 files , etc.

If attending an OGL meeting would be helpful for your work, that could also be a possibility.

There are many works in this collection where the notes were never tagged in any comprehensive, meaningful way (multiple languages, references, etc.) and this type of refinement isn't a priority at this stage.

We would welcome refinements from anyone interested.

(I also note that the credits for 1st1k work should always include the CHS and Harvard Library in addition to the other partners.)

[mgbilby]

@gcelano @lcerrato

Thank you! You're absolutely right to point out all the related impacts and implications that I hadn't considered. I was thinking only in terms of data and file management for the fairly basic structure of this simplified OGA corpus oga_src.

Thank you for the reminder about giving proper and thorough credit for data sources. I've freshly updated the readme.md to show appreciation to the many funders and contributing institutions of these rich linguistic corpora.

I'll continue to consider ways to contribute to those lower priority upstream editorial refinements that may have larger downstream impact on syntactical parsing and analysis.

I'd be delighted to attend an upcoming OGA meeting. My UVA alum email address is a reliable way to reach me.

[gregorycrane]

One of the big to-do tasks is to markup scholia, with the Homer scholia at the top of the list. Lemmas should absolutely be tagged as such and readers should not have to know up front which volume in a multi-volume edition contains scholia on Odyssey 5, for example.