def valid_desc(self, stop_metric):
print("valid desc")
valid_links = pd.DataFrame(self.kgs.valid_links)
desc1 = self.e_desc.loc[valid_links.values[:, 0]].values
desc2 = self.e_desc.loc[valid_links.values[:, 1]].values
desc_em1 = self.word_em[desc1]
desc_em2 = self.word_em[desc2]
dem1, dem2 = self.session.run(
fetches=[self.desc_embedding1, self.desc_embedding2],
feed_dict={
self.desc1: desc_em1,
self.desc2: desc_em2
})
hits1_12, mrr_12 = valid(dem1,
dem2,
None,
self.args.top_k,
self.args.test_threads_num,
metric=self.args.eval_metric,
normalize=self.args.eval_norm,
csls_k=0,
accurate=False)
# del dem1, dem2, desc_em1, desc_em2, desc1, desc2
# gc.collect()
return hits1_12 if stop_metric == 'hits1' else mrr_12
def test(model, embed_choice='avg', w=(1, 1, 1)):
if embed_choice == 'nv':
ent_embeds = model.name_embeds.eval(session=model.session)
elif embed_choice == 'rv':
ent_embeds = model.rv_ent_embeds.eval(session=model.session)
elif embed_choice == 'av':
ent_embeds = model.av_ent_embeds.eval(session=model.session)
elif embed_choice == 'final':
ent_embeds = model.ent_embeds.eval(session=model.session)
elif embed_choice == 'avg':
ent_embeds = w[0] * model.name_embeds.eval(session=model.session) + \
w[1] * model.rv_ent_embeds.eval(session=model.session) + \
w[2] * model.av_ent_embeds.eval(session=model.session)
else: # wavg
ent_embeds = model.ent_embeds
print(embed_choice, 'test results:')
embeds1 = ent_embeds[model.kgs.test_entities1, ]
embeds2 = ent_embeds[model.kgs.test_entities2, ]
hits1_12, mrr_12 = eva.valid(embeds1,
embeds2,
None,
model.args.top_k,
model.args.test_threads_num,
normalize=True)
del embeds1, embeds2
gc.collect()
return mrr_12
def valid_(self, stop_metric):
embedding = self.sess.run(self.output)
embeds1 = np.array([embedding[e] for e in self.kgs.valid_entities1])
embeds2 = np.array([embedding[e] for e in self.kgs.valid_entities2 + self.kgs.test_entities2])
hits1_12, mrr_12 = valid(embeds1, embeds2, None, self.args.top_k, self.args.test_threads_num,
metric=self.args.eval_metric)
if stop_metric == 'hits1':
return hits1_12
return mrr_12
def valid(self, stop_metric):
embeds1, embeds2, mapping = self._eval_valid_embeddings()
hits1_12, mrr_12 = valid(embeds1,
embeds2,
mapping,
self.args.top_k,
self.args.test_threads_num,
metric=self.args.eval_metric,
normalize=self.args.eval_norm,
csls_k=0,
accurate=False)
return hits1_12 if stop_metric == 'hits1' else mrr_12
def valid_WVA(model):
nv_ent_embeds1 = tf.nn.embedding_lookup(
model.name_embeds,
model.kgs.valid_entities1).eval(session=model.session)
rv_ent_embeds1 = tf.nn.embedding_lookup(
model.rv_ent_embeds,
model.kgs.valid_entities1).eval(session=model.session)
av_ent_embeds1 = tf.nn.embedding_lookup(
model.av_ent_embeds,
model.kgs.valid_entities1).eval(session=model.session)
weight11, weight21, weight31 = wva(nv_ent_embeds1, rv_ent_embeds1,
av_ent_embeds1)
test_list = model.kgs.valid_entities2 + model.kgs.test_entities2
nv_ent_embeds2 = tf.nn.embedding_lookup(
model.name_embeds, test_list).eval(session=model.session)
rv_ent_embeds2 = tf.nn.embedding_lookup(
model.rv_ent_embeds, test_list).eval(session=model.session)
av_ent_embeds2 = tf.nn.embedding_lookup(
model.av_ent_embeds, test_list).eval(session=model.session)
weight12, weight22, weight32 = wva(nv_ent_embeds2, rv_ent_embeds2,
av_ent_embeds2)
weight1 = weight11 + weight12
weight2 = weight21 + weight22
weight3 = weight31 + weight32
all_weight = weight1 + weight2 + weight3
weight1 /= all_weight
weight2 /= all_weight
weight3 /= all_weight
print('weights', weight1, weight2, weight3)
embeds1 = weight1 * nv_ent_embeds1 + \
weight2 * rv_ent_embeds1 + \
weight3 * av_ent_embeds1
embeds2 = weight1 * nv_ent_embeds2 + \
weight2 * rv_ent_embeds2 + \
weight3 * av_ent_embeds2
print('wvag valid results:')
hits1_12, mrr_12 = eva.valid(embeds1,
embeds2,
None,
model.args.top_k,
model.args.test_threads_num,
normalize=True)
del nv_ent_embeds1, rv_ent_embeds1, av_ent_embeds1
del nv_ent_embeds2, rv_ent_embeds2, av_ent_embeds2
del embeds1, embeds2
gc.collect()
return mrr_12
def valid(self):
mat = get_ent_embeds_from_attributes(self.kgs,
self.eval_attribute_embeddings(),
self.selected_attributes)
embeds1 = mat[self.kgs.valid_entities1, ]
embeds2 = mat[self.kgs.valid_entities2, ]
hits1_12, mrr_12 = evaluation.valid(embeds1,
embeds2,
None,
self.args.top_k,
self.args.test_threads_num,
metric=self.args.eval_metric)
if self.args.stop_metric == 'hits1':
return hits1_12
return mrr_12
def valid_(self, stop_metric):
se = self.session.run(self.model_se.outputs, feed_dict=self.feed_dict_se)
if self.args.test_method == "sa":
ae = self.session.run(self.model_ae.outputs, feed_dict=self.feed_dict_ae)
beta = self.args.beta
embeddings = np.concatenate([se*beta, ae*(1.0-beta)], axis=1)
else:
embeddings = se
embeds1 = np.array([embeddings[e] for e in self.kgs.valid_entities1])
embeds2 = np.array([embeddings[e] for e in self.kgs.valid_entities2 + self.kgs.test_entities2])
hits1_12, mrr_12 = valid(embeds1, embeds2, None, self.args.top_k, self.args.test_threads_num,
metric=self.args.eval_metric)
if stop_metric == 'hits1':
return hits1_12
return mrr_12
