Troubleshooting Batch Process "Add Tags from Document" with Custom Metadata

DEVONthink
1

I have a large set of PDF documents which all have a CM field called “Speaker”. I want to turn this into a tag for each document. I selected all my PDFs, clicked on “Tools->Batch Process…” menu item, chose “Add Tags from Document” from the dropdown, then added my CM field using right-click menu. The field was added as %mdspeaker% correctly.

It worked very successfully, ALMOST. Out of a few thousand PDFs, there are 67 for which the tags were not created. I have re-run the batch tool on just those 67 multiple times and still no tags are made. I can manually create and add the tags on these documents, so I think the documents and the tag names are fine. The DT log and activity window show no errors when I run the batch tool. Is there some additional logging I can enable to troubleshoot this situation?

Thanks

2

This action adds only those tags that occur in the document’s body, therefore the action Add Tags should do the trick.

3

This is covered in the Appendix > Smart Rule Events & Actions section of the built-in Help and manual…

Screenshot 2024-10-02 at 3.24.10 PM
Screenshot 2024-10-02 at 3.24.10 PM1594×1644 509 KB

4

Thank you @BLUEFROG and @cgrunenberg. That worked great!

Leads to two questions:

  1. Why did “Add Tags from Document” partially work in my case?
  2. If “Add Tags from Document” only adds tags “if they are detected” in the document, why is there a field for the user to enter anything?

There must be some more subtle differences between the two options “Add Tags” and “Add Tags from Document”. Can you add some clarifying examples perhaps?

Thanks

5

To partially answer my question, I am speculating that the “Add Tags from Document” lets you perhaps specify a bunch of tags in the UI but then only apply the tags if the tag appears in the document body as a text string.

It still doesn’t make sense though in my case. My tags are the speaker name, and the speaker name was present as a string within the document body of my 67 documents that failed to get tagged using “Add Tags from Document”.

6

I am speculating that the “Add Tags from Document” lets you perhaps specify a bunch of tags in the UI but then only apply the tags if the tag appears in the document body as a text string.

That is correct.

and the speaker name was present as a string within the document body

And how have you verified this?

7

I did not read all 67 PDFs, but several I spot-checked had the speaker name visible when viewing the PDF (e.g. “Article written by So-and-so”) and the speaker’s first and last names showed up in the document word cloud, meaning the words were in the PDF text layer. Is it possible that some OCR artifact hid the full name? E.g. the first and last name words were not adjacent in the text layer? I think some of the documents had the speaker name in bold font also.