Created
December 14, 2016 22:47
-
-
Save sughodke/05c0af92516f215347431d642436a7bb to your computer and use it in GitHub Desktop.
Determine if some text is a question
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
import nltk | |
import random | |
from pprint import pprint | |
from sklearn.pipeline import Pipeline | |
from sklearn.metrics import classification_report | |
from sklearn.feature_extraction.text import HashingVectorizer | |
posts = nltk.corpus.nps_chat.xml_posts([ | |
'10-19-20s_706posts.xml', | |
'11-08-20s_705posts.xml', | |
'11-09-20s_706posts.xml' | |
]) | |
def transform(post): | |
tokens = nltk.word_tokenize(post.text) | |
tagged_tokens = nltk.pos_tag(tokens) | |
serialized = ['_'.join(z) | |
for z in tagged_tokens] | |
text = ' '.join(serialized) | |
return text, 'Question' in post.get('class') | |
def train_classifier(): | |
pipeline = Pipeline([ | |
('vect', HashingVectorizer()), | |
('clf', SGDClassifier()) | |
]) | |
featuresets = [transform(post) for post in posts] | |
random.shuffle(featuresets) | |
size = int(len(featuresets) * .1) | |
train_set, test_set = featuresets[size:], featuresets[:size] | |
X, y = zip(*train_set) | |
pipeline.fit(X, y) | |
X, y = zip(*test_set) | |
pred = pipeline.predict(X) | |
pprint([z for z in zip(X, pred, y) | |
if z[1] != z[2]]) | |
print('accuracy %f' % pipeline.score(X, y)) | |
print(classification_report(y, pred)) | |
return pipeline | |
classifier = train_classifier() |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment