Where I can find the complete list of SpaCy Dependency Parsing labels or annotations?

Question

I have try to refer to spaCy official website https://spacy.io/api/annotation#dependency-parsing but I only got list of universal dependency relation which also on https://universaldependencies.org/u/dep/

while, when I try to parse some sentences, I also got labels or annotations that not have been listed. Such as: prep, dative, and dobj, even though that those labels can be associated with preposition for prep, direct object for dobj, and ??? for dative.

Is there any reference for me to find the complete list of SpaCy Dependency Parsing labels or annotations?

Answer 1

As stated in the SpaCy documentaion, English and German use different sets of labels than UD. For English, these are ClearNLP labels.

Answer 2

You can get an overview of all labels by browsing to the respective model page in the documentation, e.g. https://spacy.io/models/en#en_core_web_lg and expanding the section "Label scheme".

On this page, it also mentions which corpus was used to train, in this case OntoNotes 5. This is where the labelset comes from.

Further, you can run spacy.explain to try and get more information:

print(spacy.explain("prep"))

gives you

prepositional modifier