def filter_evidence(self, _):
"""
Stage 2.2 of pipeline.
Build evidence for training the classifiers, from user answers
(self.answers) and unanswered evidence (self.evidence) with last
classification score certainty over self.evidence_threshold.
"""
logger.debug(u'running filter_evidence')
evidence = Knowledge(self.answers)
n = len(evidence)
evidence.update((e, score > 0.5 and 1 or 0)
for e, score in self.evidence.items()
if certainty(score) > self.evidence_threshold
and e not in self.answers)
logger.info(u'Filtering returns {} human-built evidences and {} '
u'over-threshold evidences'.format(n,
len(evidence) - n))
return evidence
def test_certainty_unknown(self):
self.assertEqual(certainty(None), 0.5)
def test_certainty_certain_yes(self):
self.assertEqual(certainty(1), 1)
def test_certainty_uncertain(self):
self.assertEqual(certainty(0.5), 0.5)
def test_certainty_certain_no(self):
self.assertEqual(certainty(0), 1)