allenai / pawls

Software that makes labeling PDFs easy.
https://pawls.apps.allenai.org
Apache License 2.0
380 stars 74 forks source link

Do we plan to support entity annotation? #141

Open mamafun opened 3 years ago

mamafun commented 3 years ago

Sometime an entity such as person name, or address may across two lines but not fully contains the tokens of the covered line. Similar things applies to a sentence annotation. For example, currently I find no way to annotate the sentence inside the red line area. Thank you for your answers. image

codeviking commented 3 years ago

I believe there's a mechanism for relating annotations that might capture your intent here.

@DeNeutoy @lolipopshock Do either of you have any ideas here?

kacper-trebacz-atos commented 3 years ago

Quick solution would be to create a relation e.g: "same". Than you can create first entity for those first few lines of text and second for half of last line and connect them with "same" relation. Later in postprocessing you can join them. I would also like to see a dedicated solution like holding alt or ctrl while annotating entities them would be nice.

Edit: Another quick solution create entity called exclude and exclude given tokens in post processing.