Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Missing augmented dependencies of nouns controlling relative pronouns in PP #127

Open
amir-zeldes opened this issue Feb 2, 2021 · 3 comments
Assignees

Comments

@amir-zeldes
Copy link
Contributor

I think the DEPS 6:obl:with|12:nmod should read 6:obl:with|12:nmod:of in this example:

# sent_id = weblog-juancole.com_juancole_20040404101100_ENG_20040404_101100-0008
# text = Rumsfeld, Wolfowitz and Feith staffed the CPA with Neoconservatives, most of whom had no administrative experience, no Arabic, and no respect for Muslim culture (or knowledge about it).
1	Rumsfeld	Rumsfeld	PROPN	NNP	Number=Sing	6	nsubj	6:nsubj	SpaceAfter=No
2	,	,	PUNCT	,	_	3	punct	3:punct	_
3	Wolfowitz	Wolfowitz	PROPN	NNP	Number=Sing	1	conj	1:conj:and|6:nsubj	_
4	and	and	CCONJ	CC	_	5	cc	5:cc	_
5	Feith	Feith	PROPN	NNP	Number=Sing	1	conj	1:conj:and|6:nsubj	_
6	staffed	staff	VERB	VBD	Mood=Ind|Tense=Past|VerbForm=Fin	0	root	0:root	_
7	the	the	DET	DT	Definite=Def|PronType=Art	8	det	8:det	_
8	CPA	CPA	PROPN	NNP	Number=Sing	6	obj	6:obj	_
9	with	with	ADP	IN	_	10	case	10:case	_
10	Neoconservatives	neoconservative	NOUN	NNS	Number=Plur	6	obl	6:obl:with|12:nmod	SpaceAfter=No
11	,	,	PUNCT	,	_	10	punct	10:punct	_
12	most	most	ADJ	JJS	Degree=Sup	15	nsubj	15:nsubj	_
13	of	of	ADP	IN	_	14	case	14:case	_
14	whom	whom	PRON	WP	PronType=Int	12	nmod	10:ref	_
15	had	have	VERB	VBD	Mood=Ind|Tense=Past|VerbForm=Fin	10	acl:relcl	10:acl:relcl	_
16	no	no	DET	DT	_	18	det	18:det	_
17	administrative	administrative	ADJ	JJ	Degree=Pos	18	amod	18:amod	_
18	experience	experience	NOUN	NN	Number=Sing	15	obj	15:obj	SpaceAfter=No

This seems to be a recurring problem, for example in 15 here too:

# sent_id = weblog-blogspot.com_alaindewitt_20060827093500_ENG_20060827_093500-0003
# text = The United States goes into a war zone and evacuates a bunch of U.S. citizens (most of whom were "dual-citizens").
1	The	the	DET	DT	Definite=Def|PronType=Art	3	det	3:det	_
2	United	United	PROPN	NNP	Number=Sing	3	compound	3:compound	_
3	States	States	PROPN	NNP	Number=Sing	4	nsubj	4:nsubj|10:nsubj	_
4	goes	go	VERB	VBZ	Mood=Ind|Number=Sing|Person=3|Tense=Pres|VerbForm=Fin	0	root	0:root	_
5	into	into	ADP	IN	_	8	case	8:case	_
6	a	a	DET	DT	Definite=Ind|PronType=Art	8	det	8:det	_
7	war	war	NOUN	NN	Number=Sing	8	compound	8:compound	_
8	zone	zone	NOUN	NN	Number=Sing	4	obl	4:obl:into	_
9	and	and	CCONJ	CC	_	10	cc	10:cc	_
10	evacuates	evacuate	VERB	VBZ	Mood=Ind|Number=Sing|Person=3|Tense=Pres|VerbForm=Fin	4	conj	4:conj:and	_
11	a	a	DET	DT	Definite=Ind|PronType=Art	12	det	12:det	_
12	bunch	bunch	NOUN	NN	Number=Sing	10	obj	10:obj	_
13	of	of	ADP	IN	_	15	case	15:case	_
14	U.S.	U.S.	PROPN	NNP	Number=Sing	15	compound	15:compound	_
15	citizens	citizen	NOUN	NNS	Number=Plur	12	nmod	12:nmod:of|17:nmod	_
16	(	(	PUNCT	-LRB-	_	24	punct	24:punct	SpaceAfter=No
17	most	most	ADJ	JJS	Degree=Sup	24	nsubj	24:nsubj	_
18	of	of	ADP	IN	_	19	case	19:case	_
19	whom	whom	PRON	WP	PronType=Int	17	nmod	15:ref	_

Unless I'm misunderstanding the enhanced dependencies specification - if so please let me know!

@sebschu
Copy link
Member

sebschu commented Feb 3, 2021

Agreed! That looks like an issue with the enhancer. (When multiple enhancements interact, things can get quite tricky and it seems like this in an edge case with a preposition attached to the relative pronoun that is currently not handled correctly.)

We'll try to fix this before the next release.

@sebschu sebschu self-assigned this Feb 3, 2021
@amir-zeldes
Copy link
Contributor Author

Great, thanks! I've been looking at the enhancer implementation here: https://github.com/stanfordnlp/CoreNLP/blob/master/src/edu/stanford/nlp/trees/ud/UniversalEnhancer.java

Can you give an example of how to invoke it via CoreNLP? It seems to need some kind of embeddings for the gapping component, but I couldn't find documentation explaining how to run it.

@nschneid
Copy link
Contributor

nschneid commented Jan 29, 2022

The above nmod examples were fixed in #278, but there are obl edges missing prepositional refinements: http://match.grew.fr/?corpus=UD_English-EWT@dev&custom=61f4bfac8a4d5&clustering=X.upos

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

No branches or pull requests

3 participants