ontonotes coreference data

[jeswan]

Issue by lovodkin93
Monday Jul 06, 2020 at 18:20 GMT
Originally opened as nyu-mll/jiant#1102

---

Hello,
So going over the ontonotes data, I understood that the last column represents the coreference relations, as seen in the following screen shot:

as can be seen in the screenshot, two entities that are related get the same number in the last column.
Now, my question is - do these numbers have a special meaning, other than the fact the connected entities get the same ones? Or does every group of connected entitites get their index at random?
In other words, if for example I change all the "42" in all the data with a different number (that doesn't appear in the data at all) - will it affect anything?
Thanks!

[jeswan]

Comment by sleepinyourhat
Wednesday Jul 08, 2020 at 16:16 GMT

---

If you're looking at the raw data, I'm not sure we'll be able to help. You'd be better off contacting the creators of Ontonotes.

If you're specifically focused on how jiant 1.3 loads the data, @iftenney or @zphang might be able to help.