def init(self):
assert self.args.alpha > 1
self.desc_batch_size = self.args.desc_batch_size
self.negative_indication_weight = -1. / self.desc_batch_size
self.wv_dim = self.args.wv_dim
self.default_desc_length = self.args.default_desc_length
self.word_embed = self.args.word_embed
self.desc_sim_th = self.args.desc_sim_th
self.sim_th = self.args.sim_th
self.word_em, self.e_desc = self._get_desc_input()
self.ref_entities1 = self.kgs.valid_entities1 + self.kgs.test_entities1
self.ref_entities2 = self.kgs.valid_entities2 + self.kgs.test_entities2
self._define_variables()
self._define_mapping_variables()
self._define_embed_graph()
self._define_mapping_graph()
self._define_mapping_graph_new()
self._define_desc_graph()
self.session = load_session()
tf.global_variables_initializer().run(session=self.session)
def init(self):
self._define_variables()
self._define_embed_graph()
self._define_alignment_graph()
self._define_likelihood_graph()
self.session = load_session()
tf.global_variables_initializer().run(session=self.session)
self.ref_ent1 = self.kgs.valid_entities1 + self.kgs.test_entities1
self.ref_ent2 = self.kgs.valid_entities2 + self.kgs.test_entities2
# customize parameters
assert self.args.init == 'normal'
assert self.args.alignment_module == 'swapping'
assert self.args.loss == 'limited'
assert self.args.neg_sampling == 'truncated'
assert self.args.optimizer == 'Adagrad'
assert self.args.eval_metric == 'inner'
assert self.args.loss_norm == 'L2'
assert self.args.ent_l2_norm is True
assert self.args.rel_l2_norm is True
assert self.args.pos_margin >= 0.0
assert self.args.neg_margin > self.args.pos_margin
assert self.args.neg_triple_num > 1
assert self.args.truncated_epsilon > 0.0
assert self.args.learning_rate >= 0.01
def init(self):
self._define_variables()
self._define_mapping_variables()
self._define_embed_graph()
self._define_mapping_graph()
self.session = load_session()
tf.global_variables_initializer().run(session=self.session)
def init(self):
self.ref_entities1 = self.kgs.valid_entities1 + self.kgs.test_entities1
self.ref_entities2 = self.kgs.valid_entities2 + self.kgs.test_entities2
self.paths1 = generate_2steps_path(self.kgs.kg1.relation_triples_list)
self.paths2 = generate_2steps_path(self.kgs.kg2.relation_triples_list)
self._define_variables()
self._define_embed_graph()
self._define_alignment_graph()
self.session = load_session()
tf.global_variables_initializer().run(session=self.session)
# customize parameters
assert self.args.alignment_module == 'sharing'
assert self.args.init == 'normal'
assert self.args.neg_sampling == 'uniform'
assert self.args.optimizer == 'Adagrad'
assert self.args.eval_metric == 'inner'
assert self.args.loss_norm == 'L2'
assert self.args.ent_l2_norm is True
assert self.args.rel_l2_norm is True
assert self.args.margin > 0.0
assert self.args.neg_triple_num == 1
assert self.args.sim_th > 0.0
def init(self):
assert self.args.alignment_module == 'mapping'
assert self.args.neg_triple_num > 1
assert self.args.learning_rate >= 0.01
self.num_supports = self.args.support_number
self.utils = GCN_Utils(self.args, self.kgs)
self.attr = load_attr(self.kgs.entities_num, self.kgs)
self.adj, self.ae_input, self.train = self.utils.load_data(self.attr)
self.e = self.ae_input[2][0]
self.support = [self.utils.preprocess_adj(self.adj)]
self.ph_ae = {
"support": [tf.sparse_placeholder(tf.float32) for _ in range(self.args.support_number)],
"features": tf.sparse_placeholder(tf.float32),
"dropout": tf.placeholder_with_default(0., shape=()),
"num_features_nonzero": tf.placeholder_with_default(0, shape=())
}
self.ph_se = {
"support": [tf.sparse_placeholder(tf.float32) for _ in range(self.args.support_number)],
"features": tf.placeholder(tf.float32),
"dropout": tf.placeholder_with_default(0., shape=()),
"num_features_nonzero": tf.placeholder_with_default(0, shape=())
}
self.model_ae = GCN_Align_Unit(self.args, self.ph_ae, input_dim=self.ae_input[2][1],
output_dim=self.args.ae_dim, ILL=self.train,
sparse_inputs=True, featureless=False, logging=False)
self.model_se = GCN_Align_Unit(self.args, self.ph_se, input_dim=self.e, output_dim=self.args.se_dim,
ILL=self.train, sparse_inputs=False,
featureless=True, logging=False)
self.session = load_session()
tf.global_variables_initializer().run(session=self.session)
def init(self):
self.ref_entities1 = self.kgs.valid_entities1 + self.kgs.test_entities1
self.ref_entities2 = self.kgs.valid_entities2 + self.kgs.test_entities2
self._define_variables()
self._define_embed_graph()
self._define_sim_graph()
self.session = load_session()
tf.global_variables_initializer().run(session=self.session)
# customize parameters
assert self.args.alignment_module == 'sharing'
assert self.args.init == 'normal'
assert self.args.neg_sampling == 'uniform'
assert self.args.optimizer == 'Adagrad'
assert self.args.eval_metric == 'inner'
assert self.args.loss_norm == 'L2'
assert self.args.ent_l2_norm is True
assert self.args.rel_l2_norm is True
assert self.args.neg_triple_num >= 1
assert self.args.neg_alpha >= 0.0
assert self.args.top_attr_threshold > 0.0
assert self.args.attr_sim_mat_threshold > 0.0
assert self.args.attr_sim_mat_beta > 0.0
def init(self):
self.attribute_triples_list1, self.attribute_triples_list2, self.value_id_char_ids, self.char_list_size = \
formatting_attr_triples(self.kgs, self.args.literal_len)
self._define_variables()
self._define_embed_graph()
self.session = load_session()
tf.global_variables_initializer().run(session=self.session)
def init(self):
self._define_variables()
self._define_embed_graph()
self.session = load_session()
tf.global_variables_initializer().run(session=self.session)
assert self.args.init == 'xavier'
assert self.args.alignment_module == 'sharing'
assert self.args.optimizer == 'Adam'
assert self.args.eval_metric == 'inner'
assert self.args.ent_l2_norm is True
assert self.args.rel_l2_norm is True
assert self.args.dnn_neg_nums > 1
def init(self):
self.session = load_session()
tf.global_variables_initializer().run(session=self.session)
# customize parameters
assert self.args.init == 'unit'
assert self.args.alignment_module == 'mapping'
assert self.args.optimizer == 'Adagrad'
assert self.args.eval_metric == 'inner'
assert self.args.ent_l2_norm is True
assert self.args.alpha > 1
def init(self):
self.embedding_range = (self.args.gamma + self.epsilon) / self.args.dim
self._define_variables()
self._define_embed_graph()
self._define_alignment_graph()
self.session = load_session()
tf.global_variables_initializer().run(session=self.session)
self.ref_ent1 = self.kgs.valid_entities1 + self.kgs.test_entities1
self.ref_ent2 = self.kgs.valid_entities2 + self.kgs.test_entities2
# customize parameters
assert self.args.alignment_module == 'swapping'
assert self.args.neg_triple_num > 0.0
assert self.args.truncated_epsilon > 0.0
def init(self):
self.embedding_range = (self.args.gamma + self.epsilon) / self.args.dim
self._define_variables()
self._define_embed_graph()
self.session = load_session()
tf.global_variables_initializer().run(session=self.session)
assert self.args.init == 'uniform'
assert self.args.alignment_module == 'sharing'
assert self.args.neg_sampling == 'uniform'
assert self.args.optimizer == 'Adam'
assert self.args.eval_metric == 'inner'
# assert self.args.ent_l2_norm is True
# assert self.args.rel_l2_norm is True
assert self.args.gamma > 0.0
def init(self):
self._define_variables()
self._define_embed_graph()
self.session = load_session()
tf.global_variables_initializer().run(session=self.session)
assert self.args.init == 'xavier'
assert self.args.alignment_module == 'sharing'
assert self.args.neg_sampling == 'uniform'
assert self.args.optimizer == 'Adagrad'
assert self.args.eval_metric == 'inner'
assert self.args.loss_norm == 'L2'
assert self.args.ent_l2_norm is True
assert self.args.rel_l2_norm is True
assert self.args.margin > 0.0
def init(self):
self._define_variables()
self._define_embed_graph()
self.session = load_session()
tf.global_variables_initializer().run(session=self.session)
# customize parameters
assert self.args.loss == 'margin-based'
assert self.args.alignment_module == 'mapping'
assert self.args.loss == 'margin-based'
assert self.args.neg_sampling == 'uniform'
assert self.args.optimizer == 'Adam'
assert self.args.eval_metric == 'inner'
assert self.args.loss_norm == 'L2'
assert self.args.ent_l2_norm is True
assert self.args.rel_l2_norm is True
assert self.args.neg_triple_num == 1
def init(self):
self.aligned_ent_pair_set = interactive_model(self.kgs, self.args)
self._define_variables()
self._define_embed_graph()
self.session = load_session()
tf.global_variables_initializer().run(session=self.session)
# customize parameters
assert self.args.init == 'normal'
assert self.args.loss == 'margin-based'
assert self.args.neg_sampling == 'uniform'
assert self.args.optimizer == 'SGD'
assert self.args.eval_metric == 'inner'
assert self.args.loss_norm == 'L2'
assert self.args.ent_l2_norm is True
assert self.args.rel_l2_norm is True
assert self.args.neg_triple_num == 1
assert self.args.learning_rate >= 0.01
def init(self):
# Defined in AlignE
self._define_variables()
self._define_embed_graph()
# New
self._define_alignment_graph()
self._define_likelihood_graph()
# Same as AlignE
self.session = load_session(self.args.gpu)
tf.global_variables_initializer().run(session=self.session)
# Test shouldn't be here, we let BootEA overfits the datasets as the original authors does
# THIS IS COMPLETELY WRONG, however we let BootEA having this big advantage and still performing worse
# MAYBE IT'S NOT SO WRONG
self.ref_ent1 = self.kgs.valid_entities1 + self.kgs.test_entities1 + self.kgs.extra_entities1
self.ref_ent2 = self.kgs.valid_entities2 + self.kgs.test_entities2 + self.kgs.extra_entities2
self.len_valid_test = len(self.kgs.valid_entities1 +
self.kgs.test_entities1)
# Added to improve early stopping -> Not needed anymore after rollback
# self.saved_best = False
# self.best_embeds = None
# customize parameters
assert self.args.init == 'normal'
assert self.args.alignment_module == 'swapping'
assert self.args.loss == 'limited'
assert self.args.neg_sampling == 'truncated'
assert self.args.optimizer == 'Adagrad'
assert self.args.eval_metric == 'inner'
assert self.args.loss_norm == 'L2'
assert self.args.ent_l2_norm is True
assert self.args.rel_l2_norm is True
assert self.args.pos_margin >= 0.0
assert self.args.neg_margin > self.args.pos_margin
assert self.args.neg_triple_num > 1
assert self.args.truncated_epsilon > 0.0
assert self.args.learning_rate >= 0.01
def __init__(self, data, args, attr_align_model):
super().__init__(data, args, attr_align_model)
self.flag1 = -1
self.flag2 = -1
self.early_stop = False
self._define_variables()
self._define_name_view_graph()
self._define_relation_view_graph()
self._define_attribute_view_graph()
self._define_cross_kg_entity_reference_relation_view_graph()
self._define_cross_kg_entity_reference_attribute_view_graph()
self._define_cross_kg_relation_reference_graph()
self._define_cross_kg_attribute_reference_graph()
self._define_common_space_learning_graph()
self._define_space_mapping_graph()
self.session = load_session()
tf.global_variables_initializer().run(session=self.session)
def init(self):
self._options = opts = self.args
opts.data_path = opts.training_data
self.read(data_path=self._options.data_path)
sequence_datapath = '%spaths_%.1f_%.1f' % (
self._options.data_path, self._options.alpha, self._options.beta)
if not os.path.exists(sequence_datapath):
self.sample_paths()
else:
print('load existing training sequences')
self._train_data = pd.read_csv(
'%spaths_%.1f_%.1f' %
(self._options.data_path, self._options.alpha,
self._options.beta),
index_col=0)
self._define_variables()
self._define_embed_graph()
self.session = load_session()
tf.global_variables_initializer().run(session=self.session)
def __init__(self, data, args, predicate_align_model):
super().__init__(data, args, predicate_align_model)
self.out_folder = generate_out_folder(self.args.output,
self.args.training_data, '',
self.__class__.__name__)
self.flag1 = -1
self.flag2 = -1
self.early_stop = False
self._define_variables()
self._define_name_view_graph()
self._define_relation_view_graph()
self._define_attribute_view_graph()
self._define_cross_kg_entity_reference_relation_view_graph()
self._define_cross_kg_entity_reference_attribute_view_graph()
self._define_cross_kg_attribute_reference_graph()
self._define_cross_kg_relation_reference_graph()
self._define_common_space_learning_graph()
self.session = load_session()
tf.global_variables_initializer().run(session=self.session)
def init(self):
self._define_variables()
self._define_embed_graph()
self.session = load_session()
tf.global_variables_initializer().run(session=self.session)
# customize parameters -> note they are the same as in BootEA json
assert self.args.init == 'normal'
assert self.args.alignment_module == 'swapping'
assert self.args.loss == 'limited'
assert self.args.neg_sampling == 'truncated'
assert self.args.optimizer == 'Adagrad'
assert self.args.eval_metric == 'inner'
assert self.args.loss_norm == 'L2'
assert self.args.ent_l2_norm is True
assert self.args.rel_l2_norm is True
assert self.args.pos_margin >= 0.0
assert self.args.neg_margin > self.args.pos_margin
assert self.args.neg_triple_num > 1
assert self.args.truncated_epsilon > 0.0
assert self.args.learning_rate >= 0.01
