def test_find_best_matches(self): # Match "the cat" onto "the dog chases the cat" (exact fit) matches = general_matching.find_best_matches(self.cat_dmrs, self.small_dmrs) self.assertEqual(len(matches), 1) self.assertCountEqual(matches[0].nodeid_pairs, [(2, 5), (1, 4)]) self.assertCountEqual(matches[0].link_pairs, [(Link(1, 2, 'RSTR', 'H'), Link(4, 5, 'RSTR', 'H'))]) # Match "the dog chases the cat" onto "the cat chases the dog" (inexact fit) matches = general_matching.find_best_matches(self.small_dmrs, self.reverse_dmrs) self.assertEqual(len(matches), 1) self.assertCountEqual(matches[0].nodeid_pairs, [(5, 2), (4, 1), (3, 3), (2, 5), (1, 4)]) self.assertCountEqual(matches[0].link_pairs, [(Link(4, 5, 'RSTR', 'H'), Link(1, 2, 'RSTR', 'H')), (Link(1, 2, 'RSTR', 'H'), Link(4, 5, 'RSTR', 'H'))]) # No match found matches = general_matching.find_best_matches(examples_dmrs.the_mouse(), self.reverse_dmrs) self.assertIsNone(matches) # More than one match found. matches = general_matching.find_best_matches(self.cat_dmrs, self.large_dmrs) self.assertEqual(len(matches), 2) self.assertCountEqual(matches[0].nodeid_pairs, [(2, 5), (1, 4)]) self.assertCountEqual(matches[0].link_pairs, [(Link(1, 2, 'RSTR', 'H'), Link(4, 5, 'RSTR', 'H'))]) self.assertCountEqual(matches[1].nodeid_pairs, [(2, 8), (1, 7)]) self.assertCountEqual(matches[1].link_pairs, [(Link(1, 2, 'RSTR', 'H'), Link(7, 8, 'RSTR', 'H'))])
def setUp(self): self.the_cat = examples_dmrs.the_cat().convert_to( abstractSortDictDmrs(node_key=span_pred_key)) # Checks if the matching code converts to SortDictDmrs with span_pred_key self.the_cat_chases_the_dog = examples_dmrs.the_cat_chases_the_dog( ).convert_to(abstractSortDictDmrs(node_key=span_pred_key)) self.the_dog_chases_the_cat = examples_dmrs.the_dog_chases_the_cat( ).convert_to(abstractSortDictDmrs(node_key=span_pred_key)) self.the_mouse = examples_dmrs.the_mouse() \ .convert_to(abstractSortDictDmrs(node_key=span_pred_key)) self.dog_cat = examples_dmrs.dog_cat() \ .convert_to(abstractSortDictDmrs(node_key=span_pred_key))
def test_get_matching_nodeids(self): # Match "the cat" onto "the dog chases the cat" (exact fit) matches1 = aligned_matching.get_matching_nodeids(self.the_cat, self.the_dog_chases_the_cat) self.assertEqual(len(matches1), 2) self.assertCountEqual(matches1[0], [(2, 5), (1, 1)]) self.assertCountEqual(matches1[1], [(2, 5), (1, 4)]) # all_surface = True all_matches1 = aligned_matching.get_matching_nodeids(self.the_cat, self.the_dog_chases_the_cat, all_surface=True) self.assertListEqual(matches1[1], all_matches1[1]) # Extra surface nodes self.assertCountEqual(all_matches1[0], [(2, 5), (1, 1), (None, 2), (None, 3), (None, 4)]) # Match "the dog chases the cat" onto "the cat chases the dog" (inexact fit) matches2 = aligned_matching.get_matching_nodeids(self.the_dog_chases_the_cat, self.the_cat_chases_the_dog) self.assertEqual(len(matches2), 1) self.assertCountEqual(matches2[0], [(4, 4), (3, 3), (1, 1)]) all_matches2 = aligned_matching.get_matching_nodeids(self.the_dog_chases_the_cat, self.the_cat_chases_the_dog, all_surface=True) self.assertEqual(len(all_matches2), 1) self.assertCountEqual(all_matches2[0], [(4, 4), (3, 3), (1, 1), (None, 2)]) # No match found the_mouse = examples_dmrs.the_mouse() \ .convert_to(abstractSortDictDmrs(node_key=span_pred_key)) dog_cat = examples_dmrs.dog_cat() \ .convert_to(abstractSortDictDmrs(node_key=span_pred_key)) matches = aligned_matching.get_matching_nodeids(the_mouse, dog_cat) self.assertListEqual(matches, []) # Should be the same as 'the cat'. mixed_cat = ListDmrs(surface='the cat') mixed_cat.add_node(Node(nodeid=2, pred=RealPred('cat', 'n', '1'), cfrom=4, cto=7, sortinfo=InstanceSortinfo(pers='3', num='sg', ind='+'))) mixed_cat.add_node(Node(nodeid=1, pred=RealPred('the', 'q'), cfrom=0, cto=3)) mixed_cat.add_link(Link(start=1, end=2, rargname='RSTR', post='H')) mixed = aligned_matching.get_matching_nodeids(mixed_cat, self.the_dog_chases_the_cat) self.assertListEqual(mixed, matches1)
def test_find_best_matches(self): # Match "the cat" onto "the dog chases the cat" (exact fit) matches = general_matching.find_best_matches(self.cat_dmrs, self.small_dmrs) self.assertEqual(len(matches), 1) self.assertCountEqual(matches[0].nodeid_pairs, [(2, 5), (1, 4)]) self.assertCountEqual( matches[0].link_pairs, [(Link(1, 2, 'RSTR', 'H'), Link(4, 5, 'RSTR', 'H'))]) # Match "the dog chases the cat" onto "the cat chases the dog" (inexact fit) matches = general_matching.find_best_matches(self.small_dmrs, self.reverse_dmrs) self.assertEqual(len(matches), 1) self.assertCountEqual(matches[0].nodeid_pairs, [(5, 2), (4, 1), (3, 3), (2, 5), (1, 4)]) self.assertCountEqual( matches[0].link_pairs, [(Link(4, 5, 'RSTR', 'H'), Link(1, 2, 'RSTR', 'H')), (Link(1, 2, 'RSTR', 'H'), Link(4, 5, 'RSTR', 'H'))]) # No match found matches = general_matching.find_best_matches(examples_dmrs.the_mouse(), self.reverse_dmrs) self.assertIsNone(matches) # More than one match found. matches = general_matching.find_best_matches(self.cat_dmrs, self.large_dmrs) self.assertEqual(len(matches), 2) self.assertCountEqual(matches[0].nodeid_pairs, [(2, 5), (1, 4)]) self.assertCountEqual( matches[0].link_pairs, [(Link(1, 2, 'RSTR', 'H'), Link(4, 5, 'RSTR', 'H'))]) self.assertCountEqual(matches[1].nodeid_pairs, [(2, 8), (1, 7)]) self.assertCountEqual( matches[1].link_pairs, [(Link(1, 2, 'RSTR', 'H'), Link(7, 8, 'RSTR', 'H'))])