From what I can tell, the POS tagger generates the simpler universal dependencies dataset's tags, which are like NOUN, VERB, etc.
Is there a way to get the more detailed OntoNotes tags, e.g. NNP, VBZ, etc.?
rococode
All 4 comments
@rococode the included POS models were trained to universal POS tags. For fine-grained POS, you could train your own models using Flair by following the steps in the tutorial. This would be the quickest way to get this if you have a GPU setup to train models.
I can also train models here and include them with the next release since fine-grained POS tags are a feature that many people could need. However, I am still setting up my new lab (waiting for hardware orders) so it will take some time.
alanakbik
on 14 Jan 2020
Ok, thanks for the quick reply! I'll try to train my own version then.
rococode
on 14 Jan 2020
This issue has been automatically marked as stale because it has not had recent activity. It will be closed if no further activity occurs. Thank you for your contributions.
![stale[bot] picture](https://avatars.githubusercontent.com/in/1724?v=4&s=40)
stale[bot]
on 13 May 2020
This was just added to master branch and will be part of the upcoming release.
alanakbik
on 22 May 2020
Related issues
jannenev
路
3Comments
frtacoa
路
3Comments
mnishant2
路
3Comments
happypanda5
路
3Comments
inyukwo1
路
3Comments
Most helpful comment
@rococode the included POS models were trained to universal POS tags. For fine-grained POS, you could train your own models using Flair by following the steps in the tutorial. This would be the quickest way to get this if you have a GPU setup to train models.
I can also train models here and include them with the next release since fine-grained POS tags are a feature that many people could need. However, I am still setting up my new lab (waiting for hardware orders) so it will take some time.