Formulas in a separate paragraph or in stand alone tags

[panagiotis-tsolakis]

I have noticed that formulas tend to either be in a separate paragraph by themselves (even when they are part of a bigger paragraph) or be placed in a stand-alone formula tag (even when they are part of a sentence). This can create problems during the cleaning of the TEI file and its conversion to TXT.

Also, some text might be missing before or after formulas. The text from the following screenshot follows the text of the above image. The underlined text in yellow is missing from the TEI file, while the formula is again in a stand-alone tag.

[lfoppiano]

Hi @panagiotis-tsolakis thanks again. I would need to know which Grobid version you are using, and, if possible the pdf files for testing.

Thank you

[panagiotis-tsolakis]

I used Grobid 0.8.1.

Here's the pdf file:

zhang-etal-2024-quantized.pdf

[lfoppiano]

@panagiotis-tsolakis Thanks for the report.

Regarding the formula, the example is a mistake of the model because the formula should be inline embedded in the paragraph (it's part of the text), however the having equations/formulas between paragraphs is expected for equation/formula blocks, where there is a label, usually (1) or (2), etc..

In respect of the second part of the issue, this is indeed a problem because the text that you identified, is incorrectly classified as figure caption, and then, it is correctly removed from the caption, but it's tossed on the floor.
I'm going to fix this by pushing back the discarded text into the paragraph jungle. It's a high priority issue to fix.