def test_extract_system_mentions(self):
expected_spans = sorted([
spans.Span(0, 1),
spans.Span(0, 5),
spans.Span(3, 5),
spans.Span(5, 5),
spans.Span(8, 10),
spans.Span(8, 11),
spans.Span(13, 16),
spans.Span(13, 20),
spans.Span(14, 14),
spans.Span(18, 20),
spans.Span(22, 23),
spans.Span(25, 25),
spans.Span(33, 34)
])
self.assertEqual(expected_spans, [
mention.span
for mention in mention_extractor.extract_system_mentions(
self.real_document, filter_mentions=False)[1:]
])
expected_spans = sorted([
spans.Span(2, 2),
spans.Span(4, 4),
spans.Span(6, 7),
spans.Span(6, 11),
spans.Span(9, 10),
spans.Span(9, 11)
])
self.assertEqual(expected_spans, [
mention.span
for mention in mention_extractor.extract_system_mentions(
self.another_real_document, filter_mentions=False)[1:]
])
expected_spans = sorted([
spans.Span(2, 2),
spans.Span(4, 4),
spans.Span(6, 11),
spans.Span(9, 10),
spans.Span(9, 11)
])
self.assertEqual(expected_spans, [
mention.span
for mention in mention_extractor.extract_system_mentions(
self.another_real_document, filter_mentions=True)[1:]
])
def test_extract_system_mentions(self):
expected_spans = sorted([
spans.Span(0, 1),
spans.Span(0, 5),
spans.Span(3, 5),
spans.Span(5, 5),
spans.Span(8, 10),
spans.Span(8, 11),
spans.Span(13, 16),
spans.Span(13, 20),
spans.Span(14, 14),
spans.Span(18, 20),
spans.Span(22, 23),
spans.Span(25, 25),
spans.Span(33, 34)
])
self.assertEqual(expected_spans,
[mention.span for
mention in mention_extractor.extract_system_mentions(
self.real_document, filter_mentions=False)[1:]])
expected_spans = sorted([
spans.Span(2, 2),
spans.Span(4, 4),
spans.Span(6, 7),
spans.Span(6, 11),
spans.Span(9, 10),
spans.Span(9, 11)
])
self.assertEqual(expected_spans,
[mention.span for
mention in mention_extractor.extract_system_mentions(
self.another_real_document,
filter_mentions=False)[1:]])
expected_spans = sorted([
spans.Span(2, 2),
spans.Span(4, 4),
spans.Span(6, 11),
spans.Span(9, 10),
spans.Span(9, 11)
])
self.assertEqual(expected_spans,
[mention.span for
mention in mention_extractor.extract_system_mentions(
self.another_real_document,
filter_mentions=True)[1:]])
def do_coreference(self):
testing_corpus = corpora.Corpus("input", [
self.p.run_on_doc(io.StringIO(self.txt.get("0.0", tki.END)),
"input")
])
logging.info("Extracting system mentions.")
for doc in testing_corpus:
doc.system_mentions = mention_extractor.extract_system_mentions(
doc)
mention_entity_mapping, antecedent_mapping = experiments.predict(
testing_corpus, self.extractor, self.perceptron,
clusterer.all_ante)
testing_corpus.read_coref_decisions(mention_entity_mapping,
antecedent_mapping)
logging.info("Visualize")
for doc in testing_corpus:
max_id = 0
for mention in doc.system_mentions[1:]:
set_id = mention.attributes["set_id"]
if set_id:
max_id = max(set_id, max_id)
max_id += 1
doc.annotated_mentions = []
for i, mention in enumerate(doc.system_mentions[1:]):
if mention.attributes["set_id"]:
mention.attributes[
"annotated_set_id"] = mention.attributes["set_id"]
else:
mention.attributes["annotated_set_id"] = max_id + i
doc.annotated_mentions.append(mention)
ex = error_extractors.ErrorExtractor(
testing_corpus, spanning_tree_algorithms.recall_accessibility,
spanning_tree_algorithms.precision_system_output)
ex.add_system(testing_corpus)
decisions = ex.get_errors()
visualizer = visualization.Visualizer(decisions,
"input",
for_raw_input=True)
visualizer.run()
def do_coreference(self):
testing_corpus = corpora.Corpus("input", [self.p.run_on_doc(
io.StringIO(self.txt.get("0.0", tki.END)), "input")])
logging.info("Extracting system mentions.")
for doc in testing_corpus:
doc.system_mentions = mention_extractor.extract_system_mentions(doc)
mention_entity_mapping, antecedent_mapping = experiments.predict(
testing_corpus,
self.extractor,
self.perceptron,
clusterer.all_ante
)
testing_corpus.read_coref_decisions(mention_entity_mapping, antecedent_mapping)
logging.info("Visualize")
for doc in testing_corpus:
max_id = 0
for mention in doc.system_mentions[1:]:
set_id = mention.attributes["set_id"]
if set_id:
max_id = max(set_id, max_id)
max_id += 1
doc.annotated_mentions = []
for i, mention in enumerate(doc.system_mentions[1:]):
if mention.attributes["set_id"]:
mention.attributes["annotated_set_id"] = mention.attributes[
"set_id"]
else:
mention.attributes["annotated_set_id"] = max_id + i
doc.annotated_mentions.append(mention)
ex = error_extractors.ErrorExtractor(testing_corpus,
spanning_tree_algorithms.recall_accessibility,
spanning_tree_algorithms.precision_system_output)
ex.add_system(testing_corpus)
decisions = ex.get_errors()
visualizer = visualization.Visualizer(decisions, "input",
for_raw_input=True)
visualizer.run()
perceptron = import_helper.import_from_path(args.perceptron)(priors=priors,
weights=weights,
cost_scaling=0)
extractor = instance_extractors.InstanceExtractor(
import_helper.import_from_path(args.extractor), mention_features,
pairwise_features, cost_functions.null_cost, perceptron.get_labels())
logging.info("Reading in data.")
testing_corpus = corpora.Corpus.from_file(
"testing", codecs.open(args.input_filename, "r", "utf-8"))
logging.info("Extracting system mentions.")
for doc in testing_corpus:
doc.system_mentions = mention_extractor.extract_system_mentions(doc)
mention_entity_mapping, antecedent_mapping = experiments.predict(
testing_corpus, extractor, perceptron,
import_helper.import_from_path(args.clusterer))
testing_corpus.read_coref_decisions(mention_entity_mapping, antecedent_mapping)
logging.info("Write corpus to file.")
testing_corpus.write_to_file(codecs.open(args.output_filename, "w", "utf-8"))
if args.ante:
logging.info("Write antecedent decisions to file")
testing_corpus.write_antecedent_decisions_to_file(open(args.ante, "w"))
if args.gold:
def call_cort(text_blob):
mention_features = [
features.fine_type, features.gender, features.number,
features.sem_class, features.deprel, features.head_ner,
features.length, features.head, features.first, features.last,
features.preceding_token, features.next_token, features.governor,
features.ancestry
]
pairwise_features = [
features.exact_match, features.head_match, features.same_speaker,
features.alias, features.sentence_distance, features.embedding,
features.modifier, features.tokens_contained, features.head_contained,
features.token_distance
]
# todo make sure these are exact!
model_abs = '/Users/ryanpanos/Documents/code/cort_experiments/models/model-pair-train+dev.obj' #OMG evil!
# perceptron_path = 'cort.coreference.approaches.mention_ranking.RankingPerceptron'
# extractor_path = ' cort.coreference.approaches.mention_ranking.extract_substructures'
perceptron_path = 'cort.coreference.approaches.mention_ranking.RankingPerceptron'
extractor_path = ' coreference.approaches.mention_ranking.extract_substructures'
corenlp_path = '/Users/ryanpanos/Documents/code/StanfordNLP/stanford-corenlp-full-2016-10-31/' #OMG evil!
clusterer_path = 'cort.coreference.clusterer.all_ante'
# logging.info("Loading model.")
print("Loading model . ... (this takes a while) ")
priors, weights = pickle.load(open(model_abs, "rb"))
print("Model loaded.")
# perceptron = import_helper.import_from_path(perceptron_path)(
# priors=priors,
# weights=weights,
# cost_scaling=0
# )
perceptron = RankingPerceptron(priors=priors,
weights=weights,
cost_scaling=0)
extractor = instance_extractors.InstanceExtractor(
# import_helper.import_from_path(extractor_path),
extract_substructures,
mention_features,
pairwise_features,
cost_functions.null_cost,
perceptron.get_labels())
logging.info("Reading in and preprocessing data.")
p = pipeline.Pipeline(corenlp_path)
testing_corpus = p.run_on_blob("corpus", text_blob)
logging.info("Extracting system mentions.")
for doc in testing_corpus:
doc.system_mentions = mention_extractor.extract_system_mentions(doc)
mention_entity_mapping, antecedent_mapping = experiments.predict(
testing_corpus,
extractor,
perceptron,
# import_helper.import_from_path(clusterer_path)
all_ante)
testing_corpus.read_coref_decisions(mention_entity_mapping,
antecedent_mapping)
logging.info("Write output to file.")
output_ls = []
for doc in testing_corpus:
output = doc.to_simple_output()
# my_file = codecs.open(doc.identifier + "." + args.suffix, "w", "utf-8")
# my_file.write(output)
print " output: \n" + output
# my_file.close()
output_ls.append(output)
logging.info("Done.")
return
# cost_functions.null_cost,
# perceptron.get_labels()
# )
logging.info("Reading in data.")
training_corpus = corpora.Corpus.from_file("training", codecs.open(args.input_filename, "r", "utf-8"))
logging.info("Extracting system mentions.")
dummy_counter_for_train = 0
for doc in training_corpus:
if dummy_counter_for_train % 100 == 99:
logging.info("We are extracting doc " + str(dummy_counter_for_train) + ": " + doc.identifier)
dummy_counter_for_train += 1
doc.system_mentions = mention_extractor.extract_system_mentions(doc)
# logging.info("\tVerifying attributes.")
# for doc in training_corpus:
# doc.antecedent_decisions = {}
# print(doc, doc.antecedent_decisions)
# for mention in doc.system_mentions:
# if not "antecedent" in mention.attributes.keys():
# mention.attributes["antecedent"] = None
# if not "set_id" in mention.attributes.keys():
# mention.attributes["set_id"] = None
#
# logging.info("\tExtracting instances and features.")
# substructures, arc_information = extractor.extract(training_corpus)
#
