Submitter's message ActionItems:
1. Eliot will add more clarification to the example for our replacement to @copy-to
2. Kris/Robert neded to make sure a simple one without keyscopes as well as the current one, and we'll need to review those
3. Stan will reach out to WG on how much progress they've made, and what their goals are, will check in on what they're doing.
===============================================
Minutes of the OASIS DITA TC
Tuesday, 24 February 2026
Recorded by Nancy Harrison
link to agenda for this meeting:
https://github.com/oasis-tcs/dita/wiki/Previous-agendas
Attendance:
===========
Robert Anderson, Stan Doherty, Kristen Eberlein, Nancy Harrison, Scott Hudson, Bob Johnson, Eliot Kimber, Zoe Lawson, Christina Rothwell, Eric Sirois, Leroy Steinbacher, Frank Wegmann
Business
========
Regrets: Dawn Stevens
1. Minutes
17 February 2026
https://groups.oasis-open.org/discussion/dita-tc-minutes-17-february-2026-uploaded (Harrison, 23 February 2026)
Moved by Kris, 2nd Scott, approved by TC
2. Announcements
Bob; for Boston DITA Users Group, you can now access recording of last meeting. And Carlos Sevilla will speak at March meeting about using Github actions in publishing.
3. Clarification of copy-to replacement
https://groups.oasis-open.org/discussion/agenda-item-clarification-of-copy-to-replacement Eliot Kimber, 22 February
- Eliot; we need to clarify whether there's any issue with this; Robert said that there was, at DITA Europe, but I thought it was covered. We added @ to resource-id that lets you give a 'hint;' e.g., you can say 'for app, my PDF generator uses this value of appid to work'. In the simplest case, it behaves exactly like @copy-to, but it's up to the output generator to say how it works, so it's only a hint, not a filename.
- Kris; Robert, did you hear anyone being nervous?
- Robert; where we say 'it's up to the application', we don't say how it would be done. We intended for example #3 to say how it should be done, but it doesn't spell it out.
- Eliot; OK, now I understand the issue.
***ActionItem: Eliot will add more clarification to that example
- Robert; and we'll have to make sure that the example syncs with the text.
- Kris; and there's a figure in one of the examples that has placeholder text, so we need to update that as well.
- Kris; does keyscopes complicate the issue?
- Robert; we need to have an example without keyscopes as well, to make clear you don''t have to have keyscopes to make this work.
- Kris; and we may never have reviewed this topic officially.
***ActionItem: Kris/Robert neded to make sure a simple one without keyscopes as well as the current one, and we'll need to review those
- Leroy; would it make sense to say somewhere that this is -replacing @copy-to?
- Robert; that should go in the migration guide, not in the spec itself.
- Eliot; and it's specifically called out in migrating to 2.0
4. Committee notes effort (No 02-24 updates)
Committee Notes.
https://github.com/oasis-tcs/dita/wiki/DITA-TC-Committee-Notes Updated description of mission
https://groups.oasis-open.org/discussion/updated-mission-description
- Stan; no updates from last week; I need to check with folks to see if they have any updates.
***ActionItem: Stan will reach out to WG on how much progress they've made, and what their goals are, will check in on what they're doing.
5. Continued discussion of how DITA TC members are using AI
- Kris; Christina, can you fill us in on what HP is doing?
- Christina; access to copilot, and also an in-house AI tool, good for short description, and also alt-text for images, also testing Positron, testing has gone well, will be rolliong that out in March. for proof-reading and cleanup. Acrolinx.is also being used, We're using AI to generate summaries; we often have to tweak the results, but it's still useful. Using Hippo, used searches across entire doc set; it pulls info from multiple guides and provides summary plus links to all the places it got the information from. We're looking at that for production feature. Also, we're replacing regular expresions, e.g. variable names in a spreadsheet, I can tell AI to do the work, to get reg-exps coded in DITA, e.g. going from Markdown to DITA; AI is good at things that are exact, and that kind of replacement is exact.
- Kris; so HP is using Trisoft?
- Christina; we using Tridion. little diff from pure DITA. I think, as we go from 1.3 to 2.0, AI will be very valuable. Oxygen has a refactoring tool to move occurences of an element into a new element.
- Zoe; what's specific hippo?
- Christina; an internal tool that we're using.
- Kris; does it integrate with you CCMS?
- Christina; no, we have to upload PDF stuff to it.
- Leroy; so you're working on the published output, not the source...
- Christina; yes
- Nancy; what about reg exp? in source or out of source
- Leroy; what about alt-text is it looking at images?
- Christina; yes, you need to hae an image that looks like something, in terms of European accessibility standards; that way we get a good description of the image.
And we require proofreading, e.g., to let us know if a table is missing a title. Positron can do that.
- Leroy; what about AI vs. Schematron
- Christina; if they have Schematron erros, they get notified, a push; with AI, writers have to pull, and decide to do stuff. If you want to mandate something, Schematron is better.
- Scott; even with AI, you stilll need the Schematron guardrails in place.
- Christina; we know Schematron works, whereas AI might not.
- Kris; before AI, we thought of Schematron as a poor person's Acrolinx.
- Christina; we have special internal use for AI. and you could use up all your AI tokens, so you want backup.
- Bob; wrt generating alt-text. what was the quality of the generated alt-text? how does it compare to human-generated?
- Christina; always need human oversight; if it's a vague image, you might get meaningless alt-text. We have better luck with a more precise image. can give it guidelines in your prompt. a lot of trial and error. Different AI tools create different summaries. I recommend to folks that they try it; if it works for you, use it.
- Kris; accessibility and medical illustrations; drafting recs for our internal guidelines. initial results for Copilot+ChatGPT are that it does pretty decent alt-text for images, looking at image as well as context. Still working on best prompts, depends on both quality of image, and context, and prompt quality. But it's a big time-saver. Once you get into really complex images, it's more difficult. Any one using longdescref?
- Robert; being used at Oracle, but not by my team.
- Kris; anything else?
- Eric; working on product called Syndicate by Xylene. but the gold standard for what clients are doing with AI is Avelara (Mike Iantosca). I can share more next week.
- Kris; we're using alt-text for emdical illustrations, and making our content more understandable for different user reading levels.
- Christina; very useful that you can have writing set to a desired reading level.
close 11:50 am Et
-- Ms. Nancy Harrison
---------------------------------
Nancy Harrison
Principal, Infobridge Solutions
Nancy Harrison (Personal)
---------------------------------