Repository navigation
Noun chunking missing root nouns? #469
Description
When I run the noun chunker over the phrase "100 tacos with a side of rice" it returns "a side" and "rice" but not "100 tacos".
The word "tacos" has the dependency label (dep_) of "ROOT". I think the problem may be that english_noun_chunks in spacy.syntax.iterators has 'root' as lowercase. See line 5
labels = ['nsubj', 'dobj', 'nsubjpass', 'pcomp', 'pobj',
'attr', 'root']
When I add 'ROOT' to labels it works as expected and returns "100 tacos". Of course, there may be a reason that 'root' is different from 'ROOT' that I am not aware of.
Activity
honnibal commentedon Aug 9, 2016
Definitely a bug — thanks for the report.
If you have time, would you mind submitting a pull request with a test and the patch?
jmugan commentedon Aug 9, 2016
Will do!
jmugan commentedon Aug 10, 2016
I put in the change and the test, but when I try to build locally with
python setup.py build_ext --inplaceI get the error listed at http://9.media.readthedocs.io/projects/spacy/builds/3893054/Fix Issue #469: Incorrectly cased root label in noun chunk iterator
honnibal commentedon Sep 27, 2016
Broken build is sadness — I want to make sure this doesn't happen in future.
Anyway. The problem is fixed for 1.0. Thanks again for the report.
lock commentedon May 9, 2018
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.