Autotagged tags in Tags panel, Content Panel, and Read Order do not match

[cdnatadobe]

On Windows 10 with Acrobat Pro DC version 2019.012.20040 (latest):

When I autotag a document, tags in the Tags panel may appear reasonably correct, but when the Read Order structure labels and Content types do no not match the Tags.

Shown is a fresh autotag of a PDF from InDesign. In the Tags panel, the title and "United States" are both H1, but according to the Read Order tool as well as the Contents Panel, they are both Paragraphs. When I read this with NVDA, they are both paragraphs, not headings.

The document may be downloaded at

https://www.cdc.gov/drugoverdose/pdf/pubs/2018-cdc-drug-surveillance-report.pdf

All of these mismatches have to be retagged manually.

[Bevi Chagnon - PubCom.com]

Well, that's not supposed to happen!

Should have the same tags in all 3 panels.

File a bug with Adobe Acrobat's team at www.Acrobat.UserVoice.com

--Bevi Chagnon

www.PubCom.com | Specialists for accessible Publishing + Design

[a_C_student]

Well, that's interesting. I have not seen that specific issue before. Thing is though - autotagging almost never works well. I have pretty much given up on it. In my experience "tags in the Tags panel may appear reasonably correct" is rare, and even so reasonably correct is not good enough. Autotagging requires so much clean-up and corrections it is almost always easier to hand-tag from the start - if your goal is an actually accessible document, that is conforming to ISO 14289 and WCAG 2.0.

[Bevi Chagnon - PubCom.com]

We find that Autotag is a gamble: sometimes I'm awed by how well it identified the elements correctly, even complex lists, TOCs, and tables.

But then again I'm just as amazed when it fails to identify a very simple document. And why does it insist that every document's first heading is H4?!

So we give it a whirl on our remediation projects. We either remediate further from it or toss it entirely.

It's a document-by-document analysis.