def __init__(self,
kg1: KG,
kg2: KG,
train_links,
test_links,
valid_links=None,
mode='mapping',
ordered=True,
extra_entities_percentage_valid=0.0):
# BootEA: swapping (swap entities to generate extra triples), RDGCN: mapping (calibration?? -> min ||e_1-e_2||)
if mode == "sharing":
ent_ids1, ent_ids2 = generate_sharing_id(train_links,
kg1.relation_triples_set,
kg1.entities_set,
kg2.relation_triples_set,
kg2.entities_set,
ordered=ordered)
rel_ids1, rel_ids2 = generate_sharing_id([],
kg1.relation_triples_set,
kg1.relations_set,
kg2.relation_triples_set,
kg2.relations_set,
ordered=ordered)
attr_ids1, attr_ids2 = generate_sharing_id(
[],
kg1.attribute_triples_set,
kg1.attributes_set,
kg2.attribute_triples_set,
kg2.attributes_set,
ordered=ordered)
else:
# generate unique id for each entity, relation, attribute (note id for same element is different in two KGs)
ent_ids1, ent_ids2 = generate_mapping_id(kg1.relation_triples_set,
kg1.entities_set,
kg2.relation_triples_set,
kg2.entities_set,
ordered=ordered)
rel_ids1, rel_ids2 = generate_mapping_id(kg1.relation_triples_set,
kg1.relations_set,
kg2.relation_triples_set,
kg2.relations_set,
ordered=ordered)
attr_ids1, attr_ids2 = generate_mapping_id(
kg1.attribute_triples_set,
kg1.attributes_set,
kg2.attribute_triples_set,
kg2.attributes_set,
ordered=ordered)
# convert to triples (id_ent, id_rel, id_ent)
id_relation_triples1 = uris_relation_triple_2ids(
kg1.relation_triples_set, ent_ids1, rel_ids1)
id_relation_triples2 = uris_relation_triple_2ids(
kg2.relation_triples_set, ent_ids2, rel_ids2)
# convert to triples (id_ent, id_prop, literal)
id_attribute_triples1 = uris_attribute_triple_2ids(
kg1.attribute_triples_set, ent_ids1, attr_ids1)
id_attribute_triples2 = uris_attribute_triple_2ids(
kg2.attribute_triples_set, ent_ids2, attr_ids2)
self.uri_kg1 = kg1
self.uri_kg2 = kg2
# rebuild kgs using ids, add dict generated before
kg1 = KG(id_relation_triples1, id_attribute_triples1)
kg2 = KG(id_relation_triples2, id_attribute_triples2)
kg1.set_id_dict(ent_ids1, rel_ids1, attr_ids1)
kg2.set_id_dict(ent_ids2, rel_ids2, attr_ids2)
self.uri_train_links = train_links
self.uri_test_links = test_links
# convert link to (id_ent1, id_ent2)
self.train_links = uris_pair_2ids(self.uri_train_links, ent_ids1,
ent_ids2)
self.test_links = uris_pair_2ids(self.uri_test_links, ent_ids1,
ent_ids2)
# TODO: here the entities used for testing will always some from the truth only...
self.train_entities1 = [link[0] for link in self.train_links]
self.train_entities2 = [link[1] for link in self.train_links]
self.test_entities1 = [link[0] for link in self.test_links]
self.test_entities2 = [link[1] for link in self.test_links]
if mode == 'swapping':
# generate new triples by swapping (see function for detail)
sup_triples1, sup_triples2 = generate_sup_relation_triples(
self.train_links, kg1.rt_dict, kg1.hr_dict, kg2.rt_dict,
kg2.hr_dict)
# add to the KGs
kg1.add_sup_relation_triples(sup_triples1)
kg2.add_sup_relation_triples(sup_triples2)
# generate new attribute triples by adding all literals of KG1 to corresponding entity in KG2 and viceversa
sup_triples1, sup_triples2 = generate_sup_attribute_triples(
self.train_links, kg1.av_dict, kg2.av_dict)
kg1.add_sup_attribute_triples(sup_triples1)
kg2.add_sup_attribute_triples(sup_triples2)
self.kg1 = kg1
self.kg2 = kg2
self.valid_links = list()
self.valid_entities1 = list()
self.valid_entities2 = list()
# save validation links and entities (converted to ids)
if valid_links is not None:
self.uri_valid_links = valid_links
self.valid_links = uris_pair_2ids(self.uri_valid_links, ent_ids1,
ent_ids2)
self.valid_entities1 = [link[0] for link in self.valid_links]
self.valid_entities2 = [link[1] for link in self.valid_links]
self.useful_entities_list1 = self.train_entities1 + self.valid_entities1 + self.test_entities1
self.useful_entities_list2 = self.train_entities2 + self.valid_entities2 + self.test_entities2
# Save the entities which are outside the truth for both KGs
self.extra_entities1 = list(self.kg1.entities_set -
set(self.useful_entities_list1))
self.extra_entities2 = list(self.kg2.entities_set -
set(self.useful_entities_list2))
self.extra_entities_valid1 = random.sample(
self.extra_entities1,
int(len(self.extra_entities1) * extra_entities_percentage_valid))
self.extra_entities_valid2 = random.sample(
self.extra_entities2,
int(len(self.extra_entities2) * extra_entities_percentage_valid))
self.extra_entities_test1 = list(
set(self.extra_entities1) - set(self.extra_entities_valid1))
self.extra_entities_test2 = list(
set(self.extra_entities2) - set(self.extra_entities_valid2))
self.entities_num = len(self.kg1.entities_set | self.kg2.entities_set)
self.relations_num = len(self.kg1.relations_set
| self.kg2.relations_set)
self.attributes_num = len(self.kg1.attributes_set
| self.kg2.attributes_set)
def __init__(self,
kg1: KG,
kg2: KG,
train_links,
test_links,
valid_links=None,
mode='mapping',
ordered=True):
if mode == "sharing":
ent_ids1, ent_ids2 = generate_sharing_id(train_links,
kg1.relation_triples_set,
kg1.entities_set,
kg2.relation_triples_set,
kg2.entities_set,
ordered=ordered)
rel_ids1, rel_ids2 = generate_sharing_id([],
kg1.relation_triples_set,
kg1.relations_set,
kg2.relation_triples_set,
kg2.relations_set,
ordered=ordered)
attr_ids1, attr_ids2 = generate_sharing_id(
[],
kg1.attribute_triples_set,
kg1.attributes_set,
kg2.attribute_triples_set,
kg2.attributes_set,
ordered=ordered)
else:
ent_ids1, ent_ids2 = generate_mapping_id(kg1.relation_triples_set,
kg1.entities_set,
kg2.relation_triples_set,
kg2.entities_set,
ordered=ordered)
rel_ids1, rel_ids2 = generate_mapping_id(kg1.relation_triples_set,
kg1.relations_set,
kg2.relation_triples_set,
kg2.relations_set,
ordered=ordered)
attr_ids1, attr_ids2 = generate_mapping_id(
kg1.attribute_triples_set,
kg1.attributes_set,
kg2.attribute_triples_set,
kg2.attributes_set,
ordered=ordered)
id_relation_triples1 = uris_relation_triple_2ids(
kg1.relation_triples_set, ent_ids1, rel_ids1)
id_relation_triples2 = uris_relation_triple_2ids(
kg2.relation_triples_set, ent_ids2, rel_ids2)
id_attribute_triples1 = uris_attribute_triple_2ids(
kg1.attribute_triples_set, ent_ids1, attr_ids1)
id_attribute_triples2 = uris_attribute_triple_2ids(
kg2.attribute_triples_set, ent_ids2, attr_ids2)
self.uri_kg1 = kg1
self.uri_kg2 = kg2
kg1 = KG(id_relation_triples1, id_attribute_triples1)
kg2 = KG(id_relation_triples2, id_attribute_triples2)
kg1.set_id_dict(ent_ids1, rel_ids1, attr_ids1)
kg2.set_id_dict(ent_ids2, rel_ids2, attr_ids2)
self.uri_train_links = train_links
self.uri_test_links = test_links
self.train_links = uris_pair_2ids(self.uri_train_links, ent_ids1,
ent_ids2)
self.test_links = uris_pair_2ids(self.uri_test_links, ent_ids1,
ent_ids2)
self.train_entities1 = [link[0] for link in self.train_links]
self.train_entities2 = [link[1] for link in self.train_links]
self.test_entities1 = [link[0] for link in self.test_links]
self.test_entities2 = [link[1] for link in self.test_links]
if mode == 'swapping':
sup_triples1, sup_triples2 = generate_sup_relation_triples(
self.train_links, kg1.rt_dict, kg1.hr_dict, kg2.rt_dict,
kg2.hr_dict)
kg1.add_sup_relation_triples(sup_triples1)
kg2.add_sup_relation_triples(sup_triples2)
sup_triples1, sup_triples2 = generate_sup_attribute_triples(
self.train_links, kg1.av_dict, kg2.av_dict)
kg1.add_sup_attribute_triples(sup_triples1)
kg2.add_sup_attribute_triples(sup_triples2)
self.kg1 = kg1
self.kg2 = kg2
self.valid_links = list()
self.valid_entities1 = list()
self.valid_entities2 = list()
if valid_links is not None:
self.uri_valid_links = valid_links
self.valid_links = uris_pair_2ids(self.uri_valid_links, ent_ids1,
ent_ids2)
self.valid_entities1 = [link[0] for link in self.valid_links]
self.valid_entities2 = [link[1] for link in self.valid_links]
self.useful_entities_list1 = self.train_entities1 + self.valid_entities1 + self.test_entities1
self.useful_entities_list2 = self.train_entities2 + self.valid_entities2 + self.test_entities2
self.entities_num = len(self.kg1.entities_set | self.kg2.entities_set)
self.relations_num = len(self.kg1.relations_set
| self.kg2.relations_set)
self.attributes_num = len(self.kg1.attributes_set
| self.kg2.attributes_set)
def __init__(self,
kg1: KG,
kg2: KG,
train_links,
test_links,
valid_links=None,
rel_links=None,
kg_test=None,
kg_valid=None,
mode='mapping',
ordered=True,
training_data_folder=None,
train_kg='kg12'):
if mode == "sharing":
ent_ids1, ent_ids2 = generate_sharing_id(
train_links,
kg1.relation_triples_set | kg_test.relation_triples_set
| kg_valid.relation_triples_set,
kg1.entities_set | kg_test.entities_set
| kg_valid.entities_set,
kg2.relation_triples_set | kg_test.relation_triples_set
| kg_valid.relation_triples_set,
kg2.entities_set | kg_test.entities_set
| kg_valid.entities_set,
ordered=ordered)
#rel_ids1, rel_ids2 = generate_sharing_id([], kg1.relation_triples_set, kg1.relations_set,
rel_ids1, rel_ids2 = generate_sharing_id(rel_links,
kg1.relation_triples_set,
kg1.relations_set,
kg2.relation_triples_set,
kg2.relations_set,
ordered=False)
print("Relations and ids: {} {} ".format(rel_ids1, rel_ids2))
else:
ent_ids1, ent_ids2 = generate_mapping_id(
kg1.relation_triples_set | kg_test.relation_triples_set
| kg_valid.relation_triples_set,
kg1.entities_set | kg_test.entities_set
| kg_valid.entities_set,
kg2.relation_triples_set | kg_test.relation_triples_set
| kg_valid.relation_triples_set,
kg2.entities_set | kg_test.entities_set
| kg_valid.entities_set,
ordered=ordered)
rel_ids1, rel_ids2 = generate_mapping_id(kg1.relation_triples_set,
kg1.relations_set,
kg2.relation_triples_set,
kg2.relations_set,
ordered=False)
id_relation_triples1 = uris_relation_triple_2ids(
kg1.relation_triples_set, ent_ids1, rel_ids1)
id_relation_triples2 = uris_relation_triple_2ids(
kg2.relation_triples_set, ent_ids2, rel_ids2)
if train_kg == "kg2":
id_relation_triples_test = uris_relation_triple_2ids(
kg_test.relation_triples_set, ent_ids2, rel_ids2)
id_relation_triples_valid = uris_relation_triple_2ids(
kg_valid.relation_triples_set, ent_ids2, rel_ids2)
else:
id_relation_triples_test = uris_relation_triple_2ids(
kg_test.relation_triples_set, ent_ids1, rel_ids1)
id_relation_triples_valid = uris_relation_triple_2ids(
kg_valid.relation_triples_set, ent_ids1, rel_ids1)
self.uri_kg1 = kg1
self.uri_kg2 = kg2
self.uri_kg_test = kg_test
self.uri_kg_valid = kg_valid
#Build KG instances with ids
kg1 = KG(id_relation_triples1)
kg2 = KG(id_relation_triples2)
kg_test = KG(id_relation_triples_test)
kg_valid = KG(id_relation_triples_valid)
kg1.set_id_dict(ent_ids1, rel_ids1)
kg2.set_id_dict(ent_ids2, rel_ids2)
if train_kg == "kg2":
kg_test.set_id_dict(ent_ids2, rel_ids2)
kg_valid.set_id_dict(ent_ids2, rel_ids2)
self.id2rel = {v: k for k, v in rel_ids2.items()}
kg2.set_rel_classes(self.id2rel)
self.rel_classes = kg2.rel_classes
kg_valid.set_rel_classes(self.id2rel)
kg_test.set_rel_classes(self.id2rel)
else:
kg_test.set_id_dict(ent_ids1, rel_ids1)
kg_valid.set_id_dict(ent_ids1, rel_ids1)
self.id2rel = {v: k for k, v in rel_ids1.items()}
kg1.set_rel_classes(self.id2rel)
self.rel_classes = kg1.rel_classes
kg_valid.set_rel_classes(self.id2rel)
kg_test.set_rel_classes(self.id2rel)
#For filtering rank
self.set_multi_entities_dict(id_relation_triples1 +
id_relation_triples_valid +
id_relation_triples_test)
kg_valid.set_local_multi_entities_dict(self.hr_to_multi_t,
self.tr_to_multi_h)
kg_test.set_local_multi_entities_dict(self.hr_to_multi_t,
self.tr_to_multi_h)
#For alignment evaluation
self.uri_train_links = train_links
self.uri_test_links = test_links
self.train_links = uris_pair_2ids(self.uri_train_links, ent_ids1,
ent_ids2)
self.test_links = uris_pair_2ids(self.uri_test_links, ent_ids1,
ent_ids2)
self.train_entities1 = [link[0] for link in self.train_links]
self.train_entities2 = [link[1] for link in self.train_links]
self.test_entities1 = [link[0] for link in self.test_links]
self.test_entities2 = [link[1] for link in self.test_links]
self.valid_links = list()
self.valid_entities1 = list()
self.valid_entities2 = list()
if valid_links is not None:
self.uri_valid_links = valid_links
self.valid_links = uris_pair_2ids(self.uri_valid_links, ent_ids1,
ent_ids2)
self.valid_entities1 = [link[0] for link in self.valid_links]
self.valid_entities2 = [link[1] for link in self.valid_links]
if mode == 'swapping':
sup_triples1, sup_triples2 = generate_sup_relation_triples(
self.train_links, kg1.rt_dict, kg1.hr_dict, kg2.rt_dict,
kg2.hr_dict)
kg1.add_sup_relation_triples(sup_triples1)
kg2.add_sup_relation_triples(sup_triples2)
self.kg1 = kg1
self.kg2 = kg2
self.kg_test = kg_test
self.kg_valid = kg_valid
self.useful_entities_list1 = self.train_entities1 + self.valid_entities1 + self.test_entities1
self.useful_entities_list2 = self.train_entities2 + self.valid_entities2 + self.test_entities2
#
if train_kg == "kg1":
self.entities_num = len(self.kg1.entities_set
| self.kg_test.entities_set
| self.kg_valid.entities_set)
self.relations_num = len(self.kg1.relations_set
| self.kg_test.relations_set
| self.kg_valid.relations_set)
if train_kg == "kg2":
self.entities_num = len(self.kg2.entities_set
| self.kg_test.entities_set
| self.kg_valid.entities_set)
self.relations_num = len(self.kg2.relations_set
| self.kg_test.relations_set
| self.kg_valid.relations_set)
if train_kg == "kg12":
self.entities_num = len(self.kg1.entities_set
| self.kg2.entities_set
| self.kg_test.entities_set
| self.kg_valid.entities_set)
self.relations_num = len(self.kg1.relations_set
| self.kg2.relations_set
| self.kg_test.relations_set
| self.kg_valid.relations_set)
print(
"All entities_num: {}, kg1: {}, kg2: {}, kg_test: {}, kg_valid: {}"
.format(self.entities_num, len(self.kg1.entities_set),
len(self.kg2.entities_set), len(self.kg_test.entities_set),
len(self.kg_valid.entities_set)))
