"GENE", line_dict["gene_1_entity"],
[sentence.words[j] for j in line_dict["gene_1_wordidxs"]])
gene_1_mention.is_correct = line_dict["gene_1_is_correct"]
gene_1_mention.type = line_dict["gene_1_type"]
gene_2_mention = Mention(
"GENE", line_dict["gene_2_entity"],
[sentence.words[j] for j in line_dict["gene_2_wordidxs"]])
gene_2_mention.is_correct = line_dict["gene_2_is_correct"]
gene_2_mention.type = line_dict["gene_2_type"]
# If the word indexes do not overlap, create the relation candidate
# TODO there may be other cases. Check with Emily.
if not set(line_dict["gene_1_wordidxs"]) & \
set(line_dict["gene_2_wordidxs"]):
relation = Relation(
"GENEGENE", gene_1_mention, gene_2_mention)
# Add features
add_features(relation, gene_1_mention, gene_2_mention,
sentence)
# Supervise
# One of the two mentions (or both) is labelled as False
# We do not create a copy in this case because there will
# already be an unsupervised copy built on the unsupervised
# copies of the mentions.
if gene_1_mention.is_correct is False or \
gene_2_mention.is_correct is False:
relation.is_correct = False
relation.type = "GENEGENE_SUP_F"
# TODO Check in Emily's code how to supervise as True
# Print!
print(relation.tsv_dump())
gene_1_mention = Mention(
"GENE", line_dict["gene_1_entity"],
[sentence.words[j] for j in line_dict["gene_1_wordidxs"]])
gene_1_mention.is_correct = line_dict["gene_1_is_correct"]
gene_1_mention.type = line_dict["gene_1_type"]
gene_2_mention = Mention(
"GENE", line_dict["gene_2_entity"],
[sentence.words[j] for j in line_dict["gene_2_wordidxs"]])
gene_2_mention.is_correct = line_dict["gene_2_is_correct"]
gene_2_mention.type = line_dict["gene_2_type"]
# If the word indexes do not overlap, create the relation candidate
# TODO there may be other cases. Check with Emily.
if not set(line_dict["gene_1_wordidxs"]) & \
set(line_dict["gene_2_wordidxs"]):
relation = Relation("GENEGENE", gene_1_mention, gene_2_mention)
# Add features
add_features(relation, gene_1_mention, gene_2_mention,
sentence)
# Supervise
# One of the two mentions (or both) is labelled as False
# We do not create a copy in this case because there will
# already be an unsupervised copy built on the unsupervised
# copies of the mentions.
if gene_1_mention.is_correct is False or \
gene_2_mention.is_correct is False:
relation.is_correct = False
relation.type = "GENEGENE_SUP_F"
# TODO Check in Emily's code how to supervise as True
# Print!
print(relation.tsv_dump())
avail_wordidxs = list(avail_wordidxs)
if len(avail_wordidxs) > 0:
fake_rels = []
for (gene_mention, hpoterm_mention) in positive_relations:
other_word = sentence.words[random.choice(avail_wordidxs)]
fake_gene_mention = Mention(
"FAKE_GENE", other_word.lemma, [other_word, ])
fake_hpo_mention = Mention(
"FAKE_HPOTERM", other_word.lemma, [other_word, ])
fake_rel_1 = Relation(
"GENEPHENO_SUP_POSFAKEGENE", fake_gene_mention,
hpoterm_mention)
fake_rel_2 = Relation(
"GENEPHENO_SUP_POSFAKEHPO", gene_mention,
fake_hpo_mention)
fake_rel_1.is_correct = False
fake_rel_2.is_correct = False
# Print!
print(fake_rel_1.tsv_dump())
print(fake_rel_2.tsv_dump())
# Create more artificial negative examples:
# for each gene candidate G in the sentence, if the pattern G
# <Verb> X appears in the same sentence and X is not a phenotype
# mention candidate, add (gene, X) as negative examples
for gene_mention in gene_mentions:
try:
next_word = sentence.words[gene_mention.wordidxs[-1] + 1]
except IndexError:
continue
if re.search('^VB[A-Z]*$', next_word.pos) and \
next_word.word not in ["{", "}", "(", ")", "[", "]"]:
