def test_cosine_pickle(self) :
cosine = CosineSetSimilarity(self.ilist)
s1 = self.ilist[0]
s2 = self.ilist[1]
cosine_sim = cosine(s1, s2)
pickle.dumps(cosine)
cosine = CosineSetSimilarity([])
s1 = self.ilist[0]
s2 = self.ilist[1]
cosine_sim = cosine(s1, s2)
pickle.dumps(cosine)
def __init__(self, definition):
super(SetType, self).__init__(definition)
if 'corpus' not in definition:
definition['corpus'] = []
self.comparator = CosineSetSimilarity(definition['corpus'])
def test_cosine_no_corpus(self):
cosine = CosineSetSimilarity([])
s1 = self.ilist[0]
s2 = self.ilist[1]
cosine_sim = cosine(s1, s2)
self.assertAlmostEqual(cosine_sim, 0.667, places=3)
cosine_sim = cosine(('g', 'h', 'd k'), s2)
self.assertAlmostEqual(cosine_sim, 0.333, places=3)
def test_cosine_cache(self):
cosine = CosineSetSimilarity(self.ilist)
s1 = self.ilist[0]
s2 = self.ilist[1]
cosine_sim = cosine(s1, s2)
self.assertAlmostEqual(cosine_sim, 0.378, places=3)
cosine_sim = cosine(s1, s2)
self.assertAlmostEqual(cosine_sim, 0.378, places=3)
def __init__(self, definition) :
super(SetType, self).__init__(definition)
canopy_predicates = [predicates.TfidfSetPredicate(threshold,
self.field)
for threshold in self._canopy_thresholds]
self.predicates += canopy_predicates
if 'corpus' not in definition :
definition['corpus'] = []
self.comparator = CosineSetSimilarity(definition['corpus'])
def test_cosine_identical(self):
cosine = CosineSetSimilarity(self.ilist)
cosine_sim = cosine(self.ilist[0], self.ilist[0])
self.assertAlmostEqual(cosine_sim, 1, places=5)
def test_cosine_na(self):
cosine = CosineSetSimilarity(self.ilist)
cosine_sim = cosine(self.ilist[0], ())
assert numpy.isnan(cosine_sim)