def extract_labels(self, example_collection: ExampleCollection):
example_wrappers_sp = example_collection.get_example_wrappers_sp()
example_wrappers_clausedb = example_collection.get_example_wrappers_clausedb(
)
keys_of_unlabeled_examples = set()
label_distribution = {}
for ex_index, clause_db_ex in enumerate(
example_wrappers_clausedb
): # type: Tuple[int, ClauseDBExampleWrapper]
if clause_db_ex.classification_term is not None:
label = clause_db_ex.classification_term.args[
self.index_of_label_var]
self.labels.add(label)
clause_db_ex.label = label
if label in label_distribution.keys():
label_distribution[label] = label_distribution[label] + 1
else:
label_distribution[label] = 1
else:
# TODO: update this to the probabilistic way of querying
query_results = self.engine.query(clause_db_ex.logic_program,
self.predicate_to_query)
if len(query_results) is 0:
example_str = ""
for ex_statement in clause_db_ex:
print(ex_statement)
print("--")
keys_of_unlabeled_examples.add(clause_db_ex.get_key())
# raise Exception("Querying the predicate", self.predicate_to_query, "on the example gives no results. Example: \n", example_str)
for answer in query_results:
label = answer[self.index_of_label_var]
self.labels.add(label)
clause_db_ex.label = label
if label in label_distribution.keys():
label_distribution[
label] = label_distribution[label] + 1
else:
label_distribution[label] = 1
# --------------------------------
example_wrappers_sp[ex_index].label = clause_db_ex.label
# ---------------------------------------------
# set flags
example_collection.are_sp_examples_labeled = True
example_collection.are_clausedb_examples_labeled = True
nb_examples = len(example_wrappers_clausedb)
print("nb of examples: " + str(nb_examples))
nb_unlabeled_examples = len(keys_of_unlabeled_examples)
print("nb of unlabeled examples: " + str(nb_unlabeled_examples))
print("nb of labeled examples: " +
str(nb_examples - nb_unlabeled_examples))
print("indexes of unlabeled examples:")
print(keys_of_unlabeled_examples)
print("label distribution:")
print(label_distribution)
return keys_of_unlabeled_examples
def extract_labels(self, example_collection: ExampleCollection):
for example_wrapper in example_collection.get_example_wrappers_sp():
self.extract_label(example_wrapper)
example_collection.are_sp_examples_labeled = True