def test_predict_heads_with_novelties(self):
"""Test scoring heads with labeling as novel w.r.t. training and testing."""
heads_df = get_head_prediction_df(self.model, 'conferences', 'brazil', testing=self.testing_mapped_triples)
self.assertEqual(['head_id', 'head_label', 'score', 'in_training', 'in_testing'], list(heads_df.columns))
self.assertEqual(self.model.num_entities, len(heads_df.index))
training_heads = set(heads_df.loc[heads_df['in_training'], 'head_label'])
self.assertEqual({'usa', 'india', 'ussr', 'poland', 'cuba'}, training_heads)
testing_heads = set(heads_df.loc[heads_df['in_testing'], 'head_label'])
self.assertEqual(set(), testing_heads)
def test_predict_heads_with_novelties(self):
"""Test scoring heads with labeling as novel w.r.t. training and testing."""
heads_df = get_head_prediction_df(
self.model,
"conferences",
"brazil",
triples_factory=self.dataset.training,
testing=self.testing_mapped_triples,
)
self.assertEqual(
["head_id", "head_label", "score", "in_training", "in_testing"],
list(heads_df.columns))
self.assertEqual(self.model.num_entities, len(heads_df.index))
training_heads = set(heads_df.loc[heads_df["in_training"],
"head_label"])
self.assertEqual({"usa", "india", "ussr", "poland", "cuba"},
training_heads)
testing_heads = set(heads_df.loc[heads_df["in_testing"], "head_label"])
self.assertEqual(set(), testing_heads)
indices=None).cpu().detach().numpy()
e_idx = ann(entity_embedding_tensor)
e_idx.save_index(entity_index)
from pykeen.models import predict
terms = [
'http://purl.enanomapper.org/onto/ENM_8000299',
'http://semanticscience.org/resource/CHEMINF_000228',
'http://purl.jp/bio/4/id/200906029372903982',
'http://purl.enanomapper.org/onto/ENM_0000084'
]
for term in terms:
print(term)
try:
predicted_heads_df = predict.get_head_prediction_df(
result.model, 'keyword', term, triples_factory=result.training)
predicted_heads_df.head()
except:
pass
with open("{}/entity_id_to_label.json".format(kg_file), 'r') as infile:
entity_id_to_label = json.load(infile)
import pandas as pd
df = pd.DataFrame.from_dict(entity_id_to_label, orient='index', columns=["id"])
import numpy as np
from sklearn.decomposition import PCA
pca = PCA(n_components=2)
pca.fit(entity_embedding_tensor)
X = pca.transform(entity_embedding_tensor)
df["x"] = X[:, 0]