问题
from nltk.stem import WordNetLemmatizer
x = WordNetLemmatizer()
x.lemmatize("angrily", pos='r')
Out[41]: 'angrily'
Here is reference documnetation for pos tags in nltk wordnet, http://www.nltk.org/_modules/nltk/corpus/reader/wordnet.html
I may be missing some basic things. Please let me know
回答1:
Try:
>>> from nltk.corpus import wordnet as wn
>>> wn.synset('angrily.r.1').lemmas()[0].pertainyms()[0].name()
u'angry'
See Getting adjective from an adverb in nltk or other NLP library for more information
The question is why do you have to go through the lemmas to get the pertainyms?
>>> wn.synset('angrily.r.1').pertainyms()
Traceback (most recent call last):
File "<stdin>", line 1, in <module>
AttributeError: 'Synset' object has no attribute 'pertainyms'
It's because WordNet sees it as a lexical association between word categories, see http://wordnet.princeton.edu/man/wngloss.7WN.html
Pertainyms are relational adjectives and do not follow the structure just described. Pertainyms do not have antonyms; the synset for a pertainym most often contains only one word or collocation and a lexical pointer to the noun that the adjective is "pertaining to". Participial adjectives have lexical pointers to the verbs that they are derived from.
Then again, if we look at the Java interface, getting a synset's pertainym is as easy as AdjectiveSynset.getPertainyms() (http://lyle.smu.edu/~tspell/jaws/doc/edu/smu/tspell/wordnet/AdjectiveSynset.html)
So I guess it depends on who writes the interface, what sort of perspective they take towards adjective-adverb relationship.
For me I think pertainyms would have been directly related to the synset rather than the lemma.
来源：https://stackoverflow.com/questions/28475620/wordnet-lemmatizer-in-nltk-is-not-working-for-adverbs