DocEng '24 paper on tagging publications list, tagged

[nxg]

(Part (a) of this is sufficiently obvious that I imagine it's on someone's mental issue list anyway).

It would be useful (a) if the DocEng 2024 update paper (Mittelbach et al, 'Automatically producing accessible and reusable PDFs with LATEX', https://doi.org/10.1145/3685650.3685670) were included in the project publications list, and (b) if this page noted that (presumably) the papers in this list were tagged using the new support.

A link to the paper appeared on an accessible-maths list I read, and prompted some useful discussion. But one of the list participants did the obvious and sensible thing and checked to see what tagging was present in the paper, and how it read in an assistive-technologies reader. There's none, of course because (I'm guessing) the PDF was produced using a publisher workflow which (of course) doesn't include the new support; the commenter drew some negative inferences from that. Though I doubt much can be done about the proceedings version, it would be a good look if there were a canonical alternative version of the paper, here, which was an advert for itself [insert some remark about dogfood here...].

I presume it's on someone's to-do list to add the paper here, so this issue is just to log the obvious suggestion that this should be an author-prepared preprint produced with tagging, rather than the proceedings version; and that this list somehow notes that these papers have that tagging, encouraging potential readers to confirm that the PDFs illustrate the support they're describing. I suppose it would be useful to additionally indicate which version/phase of the system they're using, to avoid confusion when later phases become available.

[u-fischer]

We will include the paper in the publication list soon, and it will then be the tagged version that I created in parallel for demonstration:
doceng2024-fmi-ufi-dpc-jaw-tagged.pdf.
The tagging has not be done fully automatic. As it uses a special class, acmart, I had to add a number of patches to the preamble to activate our code. And I had to use the newest lualatex-dev. We would be interested to get feed back about it (and about the other examples). We naturally hope that the class gets adapted and that in feature everyone can create tagged PDF with this class (we spoke to the maintainer and also gave him the code we needed for our paper).

Regarding some comments in the discussion:

• Not every LaTeX document can already be tagged automatically. But a large number can. So everyone interested in producing accessible PDF should simply try and not wait until 2026 or some other year. After all, trying out only requires to add a line at the begin and look what happens (perhaps check the tagging-status of your packages first: https://latex3.github.io/tagging-project/tagging-status/).
• If there is an issue report it. We know about various open points, but we can't test everything. Also I at least tend to prioritize the issues where I know that they hinder people.
• The automatic math tagging is now activated. If you use the newest lualatex-dev and activate all tagging testphase modules in a document with unicode-math every math has a mathml associated file attached.
• The use of this mathml files naturally requires tools (PDF viewer and screen reader) that can handle them. On my machine this is currently the newest foxit reader (released today) and an experimental nvda + mathcat plugin (we will check what the status is here).
• html export has not be forgotten. At the doceng we gave a tutorial and also presented the various html export options. And in my opinion the most convincing one goes through a tagged PDF: if you pass that to a pdf-to-html derivation like it is used by ngpdf.com you get nearly on-the-fly in an easy way a very reasonable html including mathml (if AF-files are in the tagged PDF).

[nxg]

Thanks for this, Ulrike: I've passed on a pointer to your comments in the list thread.

[FrankMittelbach]

A link to the paper appeared on an accessible-maths list I read, and prompted some useful discussion.

Hi Norman, is there a simple way to join that list discussion for any of us? (I have somewhat limited internet connections right now and it is difficult to research that for me).

[car222222]

I looked at this possibility, but gave up!
So I too hope that there is an "easy way" to access elements in this truly vast collection of lists!

[nxg]

On the list archive page, I'm seeing a ‘subscribe or unsubscribe’ link in the top left, which seems to link to a subscription page, when I try it. If you're seeing something different, something might be amiss – perhaps we should resolve that offline.

[FrankMittelbach]

A PDF/UA-2 compatible version of the paper is now on the LaTeX project site at:

https://www.latex-project.org/publications/indexbytopic/pdf/