POS Mapping to UD

[chrisjbryant]

Congrats on spacy v1.0!

I just wanted to remind you that the tag_map.json to convert PTB tags to UD tags still doesn't reflect the one on the Stanford website: http://universaldependencies.org/tagset-conversion/en-penn-uposf.html
C.f. #191

I believe I've fixed your current version (attached):
tag_map.txt

The only lines that were changed:
"NIL": {"pos": "x"},
"PDT": {"pos": "det", "adjtype": "pdt", "prontype": "prn"},
"PRP$": {"pos": "det", "prontype": "prs", "poss": "poss"},
"WDT": {"pos": "det", "prontype": "int|rel"},
"WP": {"pos": "pron", "prontype": "int|rel"},
"WP$": {"pos": "det", "poss": "poss", "prontype": "int|rel"},
"AFX": {"pos": "x"}, - repeated, so deleted.
"HYPH": {"pos": "punct"}, - repeated, so deleted.

[christian-storm]

Is there a reason this is still outstanding, e.g., don't want to introduce a breaking change? Do you disagree with the UD, e.g., WDT -> DET? Spacy confusingly maps WDT -> ADJ even though the description reads 'wh-determiner'?

Happy Holidays!

[honnibal]

Sorry for not attending to this sooner. It got left out of the 2.0 changes, and then I kept glancing over the issue and not seeing that it was very simple. I think I was also worried about backwards compatibility. Will be patched in v2.1.

Fortunately we've steadily got the issue tracker under control, so these sorts of oversights shouldn't occur in future...

[lock]

This thread has been automatically locked since there has not been any recent activity after it was closed. Please open a new issue for related bugs.