Incorrect lemma capitalization

[rhdunn]

NNP should have capitalized lemma, the others lowercase:

ERROR: Sentence w01135037 token 8 -- CD/NumForm=Word lemma 'Five' does not match lowercase-form applied to form 'Five', expected 'five'
ERROR: Sentence n05001005 token 27 -- IN lemma 'Under' does not match lowercase-form applied to form 'Under', expected 'under'
ERROR: Sentence w01071043 token 3 -- NN lemma 'Post' does not match lowercase-form applied to form 'Post', expected 'post'
ERROR: Sentence w01100046 token 17 -- NN lemma 'Governor' does not match lowercase-form applied to form 'Governor', expected 'governor'
ERROR: Sentence w01100046 token 19 -- NN lemma 'General' does not match lowercase-form applied to form 'General', expected 'general'
ERROR: Sentence w01135037 token 10 -- NN lemma 'Year' does not match lowercase-form applied to form 'Year', expected 'year'
ERROR: Sentence n05001005 token 29 -- NN lemma 'Secretary' does not match lowercase-form applied to form 'Secretary', expected 'secretary'
ERROR: Sentence n02027007 token 7 -- NNP lemma 'service' does not match capitalized-form applied to form 'Service', expected 'Service'
ERROR: Sentence n02033113 token 5 -- NNP lemma 'ZEIT' does not match capitalized-form applied to form 'ZEIT', expected 'Zeit'
ERROR: Sentence n02066010 token 11 -- NNP lemma 'SPIEGEL' does not match capitalized-form applied to form 'SPIEGEL', expected 'Spiegel'
ERROR: Sentence n04006014 token 28 -- NNP lemma 'eurozone' does not match capitalized-form applied to form 'eurozone', expected 'Eurozone'
ERROR: Sentence n04008007 token 3 -- NNP lemma 'LUISS' does not match capitalized-form applied to form 'LUISS', expected 'Luiss'

[AngledLuffa]

LUISS is an acronym, I believe

https://en.wikipedia.org/wiki/Libera_Universit%C3%A0_Internazionale_degli_Studi_Sociali_Guido_Carli

Should Secret Service both be labeled NNP?

# sent_id = n02027007
# text = According to Parker, Russian Secret Service agents are active in large numbers in Great Britain.
5       Russian Russian ADJ     JJ      Degree=Pos      8       amod    8:amod  _
6       Secret  secret  NOUN    NN      Number=Sing     7       compound        7:compound      _
7       Service Service PROPN   NNP     Number=Sing     8       compound        8:compound      _

not sure eurozone should be capitalized - perhaps it's not an NNP at this point? Although it is the eurozone

https://en.wikipedia.org/wiki/Eurozone

[AngledLuffa]

still need to address up through Secretary

[nschneid]

I'm not sure "Secret Service" is the real name of any Russian entity. https://en.wikipedia.org/wiki/List_of_intelligence_agencies#Russia So I think there could be justification for NOT using NNP here.

My intuition is that "Eurozone" is specific enough to count as a proper name, even if it is sometimes lowercased.

[AngledLuffa]

I'm not sure "Secret Service" is the real name of any Russian entity.

Makes sense, but that can be confusing across countries where Secret Service is a real thing.

My intuition is that "Eurozone" is specific enough to count as a proper name, even if it is sometimes lowercased.

So leave it as NNP with a lowercase lemma?

[AngledLuffa]

Regarding Under-Secretary, I suppose the tokenization is considered correct? However, Undersecretary can also turn up as a single word. Are we happy splitting that here?

[rhdunn]

Note that with a lot of these, this is my initial assessment based on a cursory scan/check. I may have some of those wrong, like with LUISS being an initiialism/acronym.

[rhdunn]

For the hyphenated forms see also UniversalDependencies/docs#1002.

[AngledLuffa]

@rhdunn is this issue complete?