def _extractFeatures(self):
goodWordFrequencies = {
word: TextHelpers.countwordoccurences(word, self._goodText) /
self._goodWordCount
for word in {x
for x in TextHelpers.getwords(self._goodText)}
}
badWordFrequencies = {
word: TextHelpers.countwordoccurences(word, self._badText) /
self._badWordCount
for word in {x
for x in TextHelpers.getwords(self._goodText)}
}
keptWords = sorted(goodWordFrequencies.keys(),
key=(lambda k: goodWordFrequencies[k] -
badWordFrequencies.get(k, 0)),
reverse=True)[:self._featureNumber]
result = [WordClassificationFeature(x) for x in keptWords]
return result
def CountOccurrences(self, text):
return TextHelpers.countwordoccurences(self._word, text)