def align_clusters(args):
logger = Logger(args.log, args.log_specifications, sys.argv)
document_mappings = DocumentMappings(
logger, args.parent_children, Encodings(logger, args.encodings),
CoreDocuments(logger, args.core_documents))
text_boundaries = TextBoundaries(logger, args.sentence_boundaries)
image_boundaries = ImageBoundaries(logger, args.image_boundaries)
video_boundaries = VideoBoundaries(logger, args.video_boundaries)
keyframe_boundaries = KeyFrameBoundaries(logger, args.keyframe_boundaries)
document_boundaries = {
'text': text_boundaries,
'image': image_boundaries,
'keyframe': keyframe_boundaries,
'video': video_boundaries
}
annotated_regions = AnnotatedRegions(logger, document_mappings,
document_boundaries, args.regions)
os.mkdir(args.similarities)
os.mkdir(args.alignment)
for entry in sorted(os.scandir(args.gold), key=str):
if entry.is_dir() and entry.name.endswith('.ttl'):
kb = entry.name
message = 'aligning clusters in {}'.format(entry.name)
logger.record_event('DEFAULT_INFO', message)
print('At {}: {}'.format(
time.strftime("%m/%d/%Y %H:%M:%S", time.localtime()), message))
document_id = kb.replace('.ttl', '')
gold_mentions = '{}/{}/AIDA_P2_TA1_CM_A0001.rq.tsv'.format(
args.gold, kb)
gold_edges = '{}/{}/AIDA_P2_TA1_AM_A0001.rq.tsv'.format(
args.gold, kb)
system_mentions = '{}/{}/AIDA_P2_TA1_CM_A0001.rq.tsv'.format(
args.system, kb)
system_edges = '{}/{}/AIDA_P2_TA1_AM_A0001.rq.tsv'.format(
args.system, kb)
gold_mentions = gold_mentions if os.path.exists(
gold_mentions) else None
gold_edges = gold_edges if os.path.exists(gold_edges) else None
system_mentions = system_mentions if os.path.exists(
system_mentions) else None
system_edges = system_edges if os.path.exists(
system_edges) else None
similarities = '{}/{}.tab'.format(args.similarities, document_id)
alignment = '{}/{}.tab'.format(args.alignment, document_id)
check_for_paths_non_existance([similarities, alignment])
clusters = Clusters(logger, document_mappings, document_boundaries,
annotated_regions, gold_mentions, gold_edges,
system_mentions, system_edges)
clusters.print_similarities(similarities)
clusters.print_alignment(alignment)
exit(ALLOK_EXIT_CODE)
def filter_responses(args):
logger = Logger(args.log, args.log_specifications, sys.argv)
ontology_type_mappings = OntologyTypeMappings(logger,
args.ontology_type_mappings)
slot_mappings = SlotMappings(logger, args.slot_mappings)
document_mappings = DocumentMappings(
logger, args.parent_children, Encodings(logger, args.encodings),
CoreDocuments(logger, args.core_documents))
text_boundaries = TextBoundaries(logger, args.sentence_boundaries)
image_boundaries = ImageBoundaries(logger, args.image_boundaries)
video_boundaries = VideoBoundaries(logger, args.video_boundaries)
keyframe_boundaries = KeyFrameBoundaries(logger, args.keyframe_boundaries)
document_boundaries = {
'text': text_boundaries,
'image': image_boundaries,
'keyframe': keyframe_boundaries,
'video': video_boundaries
}
responses = ResponseSet(logger, ontology_type_mappings, slot_mappings,
document_mappings, document_boundaries, args.input,
args.runid)
annotated_regions = AnnotatedRegions(logger, document_mappings,
document_boundaries, args.regions)
run_filter_on_all_responses(responses, annotated_regions,
document_mappings, document_boundaries)
os.mkdir(args.output)
for input_filename in responses:
output_filename = input_filename.replace(responses.get('path'),
args.output)
dirname = os.path.dirname(output_filename)
if not os.path.exists(dirname):
os.mkdir(dirname)
output_fh = open(output_filename, 'w')
header_printed = False
for linenum in sorted(responses.get(input_filename), key=int):
entry = responses.get(input_filename).get(str(linenum))
if not header_printed:
output_fh.write('{}\n'.format(entry.get('header').get('line')))
header_printed = True
if not entry.get('valid'):
logger.record_event('EXPECTING_VALID_ENTRY',
entry.get('where'))
continue
if entry.get('passes_filter'):
output_fh.write(entry.__str__())
output_fh.close()
exit(ALLOK_EXIT_CODE)
def score_submission(args):
logger = Logger(args.log, args.log_specifications, sys.argv)
ontology_type_mappings = OntologyTypeMappings(logger,
args.ontology_type_mappings)
slot_mappings = SlotMappings(logger, args.slot_mappings)
document_mappings = DocumentMappings(
logger, args.parent_children, Encodings(logger, args.encodings),
CoreDocuments(logger, args.core_documents))
text_boundaries = TextBoundaries(logger, args.sentence_boundaries)
image_boundaries = ImageBoundaries(logger, args.image_boundaries)
video_boundaries = VideoBoundaries(logger, args.video_boundaries)
keyframe_boundaries = KeyFrameBoundaries(logger, args.keyframe_boundaries)
document_boundaries = {
'text': text_boundaries,
'image': image_boundaries,
'keyframe': keyframe_boundaries,
'video': video_boundaries
}
annotated_regions = AnnotatedRegions(logger, ontology_type_mappings,
document_mappings,
document_boundaries, args.regions)
gold_responses = ResponseSet(logger, ontology_type_mappings, slot_mappings,
document_mappings, document_boundaries,
args.gold, 'gold')
system_responses = ResponseSet(logger, ontology_type_mappings,
slot_mappings, document_mappings,
document_boundaries, args.system,
args.runid)
cluster_alignment = ClusterAlignment(logger, args.alignment)
cluster_self_similarities = ClusterSelfSimilarities(
logger, args.similarities)
scores = ScoresManager(logger, annotated_regions, gold_responses,
system_responses, cluster_alignment,
cluster_self_similarities, args.separator)
scores.print_scores(args.scores)
exit(ALLOK_EXIT_CODE)