def input_edge(self, edge):
if (not edge.is_atom() and
len(edge) == 2 and
edge[0].is_atom() and
edge[0].root() == 'the' and
has_proper_concept(edge[1])):
make_corefs(self.hg, edge, edge[1])
self.corefs += 1
def test_connect_coref_sets(self):
concepts = self.concepts
# paris set
make_corefs(self.hg, concepts[0], concepts[1])
make_corefs(self.hg, concepts[1], concepts[2])
# berlin set
make_corefs(self.hg, concepts[3], concepts[4])
make_corefs(self.hg, concepts[4], concepts[5])
self.assertTrue(are_corefs(self.hg, concepts[0], concepts[1]))
self.assertTrue(are_corefs(self.hg, concepts[0], concepts[2]))
self.assertEqual(coref_id(self.hg, concepts[0]),
coref_id(self.hg, concepts[1]))
self.assertEqual(coref_id(self.hg, concepts[0]),
coref_id(self.hg, concepts[2]))
self.assertIsNotNone(coref_id(self.hg, concepts[0]))
self.assertIsNotNone(coref_id(self.hg, concepts[1]))
self.assertIsNotNone(coref_id(self.hg, concepts[2]))
self.assertEqual(coref_set(self.hg, concepts[0]),
{concepts[0], concepts[1], concepts[2]})
self.assertTrue(are_corefs(self.hg, concepts[3], concepts[4]))
self.assertTrue(are_corefs(self.hg, concepts[3], concepts[5]))
self.assertEqual(coref_id(self.hg, concepts[3]),
coref_id(self.hg, concepts[4]))
self.assertEqual(coref_id(self.hg, concepts[3]),
coref_id(self.hg, concepts[5]))
self.assertIsNotNone(coref_id(self.hg, concepts[3]))
self.assertIsNotNone(coref_id(self.hg, concepts[4]))
self.assertIsNotNone(coref_id(self.hg, concepts[5]))
self.assertEqual(coref_set(self.hg, concepts[3]),
{concepts[3], concepts[4], concepts[5]})
self.assertFalse(are_corefs(self.hg, concepts[0], concepts[4]))
self.assertFalse(are_corefs(self.hg, concepts[1], concepts[5]))
self.assertNotEqual(coref_id(self.hg, concepts[0]),
coref_id(self.hg, concepts[4]))
self.assertNotEqual(coref_id(self.hg, concepts[1]),
coref_id(self.hg, concepts[5]))
# connect both
make_corefs(self.hg, concepts[0], concepts[5])
self.assertTrue(are_corefs(self.hg, concepts[0], concepts[4]))
self.assertTrue(are_corefs(self.hg, concepts[1], concepts[5]))
self.assertEqual(coref_id(self.hg, concepts[0]),
coref_id(self.hg, concepts[4]))
self.assertEqual(coref_id(self.hg, concepts[1]),
coref_id(self.hg, concepts[5]))
for concept in concepts:
self.assertEqual(coref_set(self.hg, concept), set(concepts))
def test_make_2_corefs(self):
concepts = self.concepts
make_corefs(self.hg, concepts[0], concepts[1])
self.assertTrue(are_corefs(self.hg, concepts[0], concepts[1]))
self.assertFalse(are_corefs(self.hg, concepts[0], concepts[2]))
self.assertEqual(coref_id(self.hg, concepts[0]),
coref_id(self.hg, concepts[1]))
self.assertIsNotNone(coref_id(self.hg, concepts[0]))
self.assertIsNone(coref_id(self.hg, concepts[2]))
self.assertEqual(coref_set(self.hg, concepts[0]),
{concepts[0], concepts[1]})
def test_main_coref(self):
concepts = self.concepts
self.assertEqual(main_coref(self.hg, concepts[0]), concepts[0])
self.assertEqual(main_coref(self.hg, concepts[1]), concepts[1])
self.assertEqual(main_coref(self.hg, concepts[2]), concepts[2])
make_corefs(self.hg, concepts[0], concepts[1])
make_corefs(self.hg, concepts[1], concepts[2])
self.assertEqual(main_coref(self.hg, concepts[0]), concepts[1])
self.assertEqual(main_coref(self.hg, concepts[1]), concepts[1])
self.assertEqual(main_coref(self.hg, concepts[2]), concepts[1])
def input_edge(self, edge):
if edge.type()[0] == 'c':
subs = tuple(subtypes(self.hg, edge))
# check if the concept should be assigned to a synonym set
if len(subs) > 0:
# find set with the highest degree and normalize set
# degrees by total degree
sub_degs = [self.hg.degree(sub) for sub in subs]
total_deg = sum(sub_degs)
total_deg = 1 if total_deg == 0 else total_deg
sub_ratios = [sub_deg / total_deg for sub_deg in sub_degs]
max_ratio = 0.
best_pos = -1
for pos, ratio in enumerate(sub_ratios):
if ratio > max_ratio:
max_ratio = ratio
best_pos = pos
# compute some degree-related metrics
sdd = self.hg.deep_degree(subs[best_pos])
_, rdd = self.hg.root_degrees(edge)
sub_to_root_dd = \
0. if rdd == 0 else float(sdd) / float(rdd)
d = self.hg.degree(edge)
dd = self.hg.deep_degree(edge)
r = float(d) / float(dd)
ld, ldd = self.hg.lemma_degrees(edge)
lr = float(ld) / float(ldd)
# use metric to decide
if (rdd > 5 and max_ratio >= .7 and r >= .05 and
lr >= .05 and sub_to_root_dd >= .1 and
(not edge.is_atom() or len(edge.root()) > 2)):
make_corefs(self.hg, edge, subs[best_pos])
self.corefs += 1
def input_edge(self, edge):
uedge = unidecode_edge(edge)
if uedge != edge and self.hg.exists(uedge):
make_corefs(self.hg, edge, uedge)
self.corefs += 1