class TestClassifierSentenceMulti(unittest.TestCase):
def setUp(self):
self.classifier = Classifier(granularity='sentence', binary=False)
def test_predict_for_groups(self):
data = [
('Its', 'it', 'PRP$', 'B-NP'), ('short', 'short', 'JJ', 'I-NP'),
('life', 'life', 'NN', 'I-NP'), ('span', 'span', 'NN', 'I-NP'),
('(', '(', '(', 'O'), ('thirty-odd', 'thirty-odd', 'JJ', 'B-NP'),
('episodes', 'episod', 'NNS', 'I-NP'), (')', ')', ')', 'O'),
('was', 'wa', 'VBD', 'B-VP'), ('in', 'in', 'IN', 'B-PP'),
('part', 'part', 'NN', 'B-NP'),
('because', 'becaus', 'IN', 'B-PP'), ('it', 'it', 'PRP', 'B-NP'),
('was', 'wa', 'VBD', 'B-VP'),
('considered', 'consid', 'VBN', 'I-VP'), ('too', 'too', 'RB', 'O'),
('violent', 'violent', 'JJ', 'B-NP'), ('at', 'at', 'IN', 'B-PP'),
('the', 'the', 'DT', 'B-NP'), ('time', 'time', 'NN', 'I-NP'),
(',', ',', ',', 'O'), ('although', 'although', 'IN', 'O'),
('the', 'the', 'DT', 'B-NP'), ('violence', 'violenc',
'NN', 'I-NP'),
('it', 'it', 'PRP', 'B-NP'), ('depicted', 'depict', 'VBD', 'B-VP'),
('is', 'is', 'VBZ', 'I-VP'), ('actually', 'actual', 'RB', 'O'),
('mild', 'mild', 'VBN', 'O'), ('by', 'by', 'IN', 'O'),
('today', 'today', 'NN', 'B-NP'), ("'s", "'s", 'POS', 'B-NP'),
('television', 'televis', 'NN', 'I-NP'),
('standards', 'standard', 'NNS', 'I-NP'), ('.', '.', '.', 'O')
]
expected = 'D'
actual = self.classifier.predict(data)
self.assertEqual(expected, actual)
def test_predict_for_sentence(self):
data = 'Its short life span (thirty-odd episodes) was in part ' \
'because it was considered too violent at the time, ' \
'although the violence it depicted is actually mild by ' \
'today\'s television standards.'
expected = 'D'
actual = self.classifier.predict(data)
self.assertEqual(expected, actual)
data = 'If I\'m human, then this test will pass without failure.'
expected = 'N'
actual = self.classifier.predict(data)
self.assertEqual(expected, actual)
data = 'I am certain that this sentence will be certain.'
expected = 'C'
actual = self.classifier.predict(data)
self.assertEqual(expected, actual)
def train(args):
classifier = Classifier(binary=not args.multiclass)
classifier.train(args.filepath)
def predict(args):
classifier = Classifier(granularity=args.granularity,
binary=not args.multiclass)
print('{}: {}'.format(args.sentence, classifier.predict(args.sentence)))
def uncertainty_features(text):
cls = Classifier()
results = cls.predict(text)
unc_ratio = sum([1 for r in results if r == 'U']) / len(results)
return unc_ratio
def setUp(self):
self.classifier = Classifier(granularity='word', binary=True)
class TestClassifierWordBinary(unittest.TestCase):
def setUp(self):
self.classifier = Classifier(granularity='word', binary=True)
def test_predict_for_groups(self):
data = [
('Its', 'it', 'PRP$', 'B-NP'), ('short', 'short', 'JJ', 'I-NP'),
('life', 'life', 'NN', 'I-NP'), ('span', 'span', 'NN', 'I-NP'),
('(', '(', '(', 'O'), ('thirty-odd', 'thirty-odd', 'JJ', 'B-NP'),
('episodes', 'episod', 'NNS', 'I-NP'), (')', ')', ')', 'O'),
('was', 'wa', 'VBD', 'B-VP'), ('in', 'in', 'IN', 'B-PP'),
('part', 'part', 'NN', 'B-NP'),
('because', 'becaus', 'IN', 'B-PP'), ('it', 'it', 'PRP', 'B-NP'),
('was', 'wa', 'VBD', 'B-VP'),
('considered', 'consid', 'VBN', 'I-VP'), ('too', 'too', 'RB', 'O'),
('violent', 'violent', 'JJ', 'B-NP'), ('at', 'at', 'IN', 'B-PP'),
('the', 'the', 'DT', 'B-NP'), ('time', 'time', 'NN', 'I-NP'),
(',', ',', ',', 'O'), ('although', 'although', 'IN', 'O'),
('the', 'the', 'DT', 'B-NP'), ('violence', 'violenc',
'NN', 'I-NP'),
('it', 'it', 'PRP', 'B-NP'), ('depicted', 'depict', 'VBD', 'B-VP'),
('is', 'is', 'VBZ', 'I-VP'), ('actually', 'actual', 'RB', 'O'),
('mild', 'mild', 'VBN', 'O'), ('by', 'by', 'IN', 'O'),
('today', 'today', 'NN', 'B-NP'), ("'s", "'s", 'POS', 'B-NP'),
('television', 'televis', 'NN', 'I-NP'),
('standards', 'standard', 'NNS', 'I-NP'), ('.', '.', '.', 'O')
]
expected = [
'C', # Its
'C', # short
'C', # life
'C', # span
'C', # (
'C', # thirty-odd
'C', # episodes
'C', # )
'C', # was
'C', # in
'C', # part
'C', # because
'C', # it
'C', # was
'U', # considered
'C', # too
'C', # violent
'C', # at
'C', # the
'C', # time
'C', # ,
'C', # although
'C', # the
'C', # violence
'C', # it
'C', # depicted
'C', # is
'C', # actually
'C', # mild
'C', # by
'C', # today
'C', # 's
'C', # television
'C', # standards
'C' # .
]
actual = self.classifier.predict(data)
self.assertEqual(expected, actual)
def test_predict_for_sentence(self):
data = 'Its short life span (thirty-odd episodes) was in part ' \
'because it was considered too violent at the time, ' \
'although the violence it depicted is actually mild by ' \
'today\'s television standards.'
expected = [
'C', # Its
'C', # short
'C', # life
'C', # span
'C', # (
'C', # thirty-odd
'C', # episodes
'C', # )
'C', # was
'C', # in
'C', # part
'C', # because
'C', # it
'C', # was
'U', # considered
'C', # too
'C', # violent
'C', # at
'C', # the
'C', # time
'C', # ,
'C', # although
'C', # the
'C', # violence
'C', # it
'C', # depicted
'C', # is
'C', # actually
'C', # mild
'C', # by
'C', # today
'C', # 's
'C', # television
'C', # standards
'C' # .
]
actual = self.classifier.predict(data)
self.assertEqual(expected, actual)
def setUp(self):
self.classifier = Classifier(granularity='sentence', binary=False)
import re
import sys
import subprocess
import traceback
import warnings
warnings.filterwarnings("ignore", category=FutureWarning)
from json import JSONDecodeError
from app.lib.helpers import JSON_NULL
from app.lib.nlp import analyzers
from uncertainty.classifier import Classifier
classifier = Classifier(granularity='word', binary=False)
class UncertaintyAnalyzer(analyzers.Analyzer):
def __init__(self, tokens, root_type='stem'):
super(UncertaintyAnalyzer, self).__init__("")
self.tokens = tokens
self.root = root_type
def analyze(self):
uncertainty = None
try:
tok_list = []
if self.root == "stem":
for tok in self.tokens: