Closed yhamoudi closed 9 years ago
Here is the list of words given by NLTK, with their probability:
[('printer', 0.25),
('imprint', 0.08333333333333333),
('impressment', 0.08333333333333333),
('impression', 0.08333333333333333),
('printing', 0.08333333333333333),
('shanghaier', 0.08333333333333333),
('instilling', 0.08333333333333333),
('impress', 0.08333333333333333),
('affect', 0.08333333333333333),
('print', 0.08333333333333333)]
This is sad, because the verb impress
means "affect strongly " whereas the noun impress
is more related to printing (see wiktionary). We could hope a better nounification for this one (without hardcoded exception).
if we hardcode non-general words (!= questions words, be, have, ...) whe should put the map out of the file where algos are
Yes. And do the same for dependenciesMap.
I'm not sure it's the same thing. dependenciesMap is a core of the algo. Its size will not increase (contrary to nounifyException that already contains non-usual words). Moreover, dependenciesMap is "special" because certain words are mapped to functions, i'm not sure it's easy to put the map out of the file that contains these functions.
Yes I agree.
What did John Hinckley do to impress Jodie Foster?
Impress >> printer :o