def get_all_expanded_types(self):
"""
For all the types asserted for the cluster, return a union
of corresponding expanded types.
"""
types = []
for cluster_type in self.get('types'):
expanded_types = get_expanded_types(self.get('metatype'), cluster_type)
types.extend(expanded_types)
return list({cluster_type:1 for cluster_type in types}.keys())
def run_filter_on_entry(entry, schema_name, filtered_clusters, annotated_regions, document_mappings, document_boundaries):
logger = entry.get('logger')
if schema_name not in ['AIDA_PHASE2_TASK1_AM_RESPONSE', 'AIDA_PHASE2_TASK1_CM_RESPONSE', 'AIDA_PHASE2_TASK1_TM_RESPONSE']:
logger.record_event('DEFAULT_CRITICAL_ERROR', 'Unexpected schema name: {}'.format(schema_name), logger.get('code_location'))
passes_filter = False
if schema_name == 'AIDA_PHASE2_TASK1_AM_RESPONSE':
cluster_type_to_keys = {
'subject': '{}:{}'.format(entry.get('kb_document_id'), entry.get('subject_cluster').get('ID')),
'object': '{}:{}'.format(entry.get('kb_document_id'), entry.get('object_cluster').get('ID'))
}
passes_filter = True
for cluster_type in cluster_type_to_keys:
key = cluster_type_to_keys[cluster_type]
if key in filtered_clusters:
if not filtered_clusters[key]:
passes_filter = False
cluster_id = entry.get('{}_cluster'.format(cluster_type)).get('ID')
logger.record_event('DEFAULT_INFO', 'Entry fails the filter due to cluster: {}'.format(cluster_id), entry.get('where'))
break
else:
logger.record_event('MISSING_ENTRY_IN_LOOKUP_ERROR', key, 'filtered_clusters', logger.get('code_location'))
elif schema_name == 'AIDA_PHASE2_TASK1_CM_RESPONSE':
mention = get_mention(entry, document_mappings, document_boundaries)
key = '{}:{}'.format(entry.get('kb_document_id'), entry.get('cluster').get('ID'))
if key not in filtered_clusters:
filtered_clusters[key] = False
if annotated_regions.contains(mention, get_expanded_types(entry.get('metatype'), entry.get('cluster_type'))):
passes_filter = True
filtered_clusters[key] = True
else:
logger.record_event('DEFAULT_INFO', 'Entry fails the filter due to mention: {}'.format(mention.get('span_string')), entry.get('where'))
elif schema_name == 'AIDA_PHASE2_TASK1_TM_RESPONSE':
cluster_id = entry.get('cluster').get('ID')
key = '{}:{}'.format(entry.get('kb_document_id'), cluster_id)
if key in filtered_clusters:
passes_filter = filtered_clusters[key]
if not passes_filter:
logger.record_event('DEFAULT_INFO', 'Entry fails the filter due to cluster: {}'.format(cluster_id), entry.get('where'))
else:
logger.record_event('MISSING_ENTRY_IN_LOOKUP_ERROR', key, 'filtered_clusters', logger.get('code_location'))
entry.set('passes_filter', passes_filter)