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)
def validate_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)
responses.write_valid_responses(args.output)
exit(ALLOK_EXIT_CODE)
def validate_responses(args):
logger = Logger(args.log, args.log_specifications, sys.argv)
logger.record_event('DEFAULT_INFO', 'validation started')
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
}
queries = TA3QuerySet(logger, args.queries) if args.queries else None
responses = ResponseSet(logger,
document_mappings,
document_boundaries,
args.input,
args.runid,
'task3',
queries=queries)
arf = AssessorReadableFormat(logger, responses, args.max_qnode_types,
args.lenient)
arf.write_output(args.output)
num_warnings, num_errors = logger.get_stats()
closing_message = 'ARF generation finished (warnings:{}, errors:{})'.format(
num_warnings, num_errors)
logger.record_event('DEFAULT_INFO', closing_message)
print(closing_message)
if num_errors > 0:
exit(ERROR_EXIT_CODE)
exit(ALLOK_EXIT_CODE)
def main(args):
logger = Logger(args.log, args.log_specifications, sys.argv)
type_mappings = Container(logger)
for entry in FileHandler(logger, args.ontology_type_mappings):
type_mappings.add(key=entry.get('full_type_ov'),
value=entry.get('full_type'))
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
}
output = []
for entry in FileHandler(logger, args.input):
document_id = entry.get('root_doc_id')
document_element_id = entry.get('doc_element_id')
modality = entry.get('media_type')
type = entry.get('type')
subtype = entry.get('subtype')
subsubtype = entry.get('subsubtype')
full_type = '{type}.{subtype}.{subsubtype}'.format(
type=type, subtype=subtype, subsubtype=subsubtype)
full_type_cleaned = full_type.replace('.unspecified', '')
propercased_full_type = type_mappings.get(full_type_cleaned, None)
span_string = entry.get('span')
keyframe_id = None
keyframe_num = 0
if span_string == 'ENTIRE_DOCUMENT_ELEMENT':
document_boundary = document_boundaries.get(modality).get(
document_element_id)
span_string = document_boundary.__str__()
elif '-' in span_string:
start, end = span_string.split('-')
span_string = '({start},0)-({end},0)'.format(start=start, end=end)
elif '_' in span_string:
keyframe_id = span_string
keyframe_num = span_string.split('_')[1]
document_boundary = document_boundaries.get('keyframe').get(
keyframe_id)
span_string = document_boundary.__str__()
else:
span_string = None
output_object = {
'document_id': document_id,
'document_element_id': document_element_id,
'keyframe_id': keyframe_id,
'keyframe_num': int(keyframe_num),
'modality': modality,
'region': span_string,
'type': propercased_full_type,
}
output.append(output_object)
printed = {}
fh = open(args.output, 'w')
header = [
'document_id', 'document_element_or_keyframe_id', 'modality', 'region',
'type'
]
fh.write('{}\n'.format('\t'.join(header)))
for output_object in multisort(
output, (('document_id', False), ('modality', False),
('document_element_id', False), ('keyframe_num', False),
('region', False), ('type', False))):
line = get_line(output_object, header)
if line not in printed:
fh.write('{}\n'.format(line))
printed[line] = 1
fh.close()
exit(ALLOK_EXIT_CODE)
def align_clusters(args):
logger = Logger(args.log, args.log_specifications, sys.argv)
ontology_type_mappings = OntologyTypeMappings(logger,
args.ontology_type_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)
thresholds = {
'ENG': args.eng_iou_threshold,
'SPA': args.spa_iou_threshold,
'RUS': args.rus_iou_threshold,
'image': args.image_iou_threshold,
'video': args.video_iou_threshold
}
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
document_id = kb.replace('.ttl', '')
if not document_mappings.get('documents').get(document_id).get(
'is_core'):
continue
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))
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, thresholds)
clusters.print_similarities(similarities)
clusters.print_alignment(alignment)
exit(ALLOK_EXIT_CODE)