def exercise3(corpus, n, out):
wordLists = getWordListsFromXForm(corpus, reviewAuthorXForm)
freqWords = review_features.freqWordList(wordLists)
features = [
review_features.createContainsFeature(freqWords, 'freq'),
review_features.distinctWordsFeature,
review_features.mostOccurringWordFeature,
review_features.numWordsFeature,
review_features.numNegationsFeature,
]
classifier = doExercise(corpus, reviewAuthorXForm, nltk.NaiveBayesClassifier.train, features, binaryrms, n, out)
return (reviewAuthorXForm, features, classifier)
def exercise4(corpus):
wordLists = getWordListsFromXForm(corpus, reviewAuthorXForm)
freqWords = review_features.freqWordList(wordLists)
features = [
review_features.createContainsFeature(freqWords, 'freq'),
review_features.distinctWordsFeature,
review_features.mostOccurringWordFeature,
review_features.numWordsFeature,
review_features.numNegationsFeature,
]
classifier = doExercise(corpus, reviewAuthorXForm, nltk.NaiveBayesClassifier.train, features)
matrix = confusion_matrix.initMatrix(list(set([review.getAuthorName() for review in corpus])))
for review in corpus:
auth = review.getAuthorName()
pAuth = classifier.classify(toFeatureSetDatum(review, reviewAuthorXForm, features)[0])
confusion_matrix.keepScore(pAuth, auth, matrix)
confusion_matrix.drawMatrix(matrix, 30)
return (reviewAuthorXForm, features, classifier)