def process(self, **kwargs):
self.log('Going to train field detector model based on the datasets stored in DB...')
document_class = kwargs.get('document_class')
task_id = kwargs.get('task_id')
if document_class:
TrainFieldDetectorModel.train_model_for_document_class(
document_class, task_id)
else:
for document_class, fields in DOCUMENT_FIELDS.items():
TrainFieldDetectorModel.train_model_for_document_class(
document_class, task_id)
def process(self, **kwargs):
self.log_info('Going to train field detector model based on the datasets stored in DB...')
document_class = kwargs.get('document_class')
if document_class:
TrainFieldDetectorModel.local_train_model_for_document_class(self, document_class)
else:
train_model_for_document_class_args = []
for document_class, fields in DOCUMENT_FIELDS.items():
train_model_for_document_class_args.append((document_class,))
self.run_sub_tasks('Train Model For Each Document Class',
TrainFieldDetectorModel.train_model_for_document_class,
train_model_for_document_class_args)
def process(self, **kwargs):
self.log_info(
"Going to prepare datasets based on the pre-coded regexps and annotations"
"entered by users...")
document_class = kwargs.get('document_class')
document_ids = kwargs.get('document_ids')
if document_class:
self.build_sentences_to_fields_relations_dataset(
document_class, document_ids)
else:
for document_class, fields in DOCUMENT_FIELDS.items():
self.build_sentences_to_fields_relations_dataset(
document_class, document_ids)
def process(self, **kwargs):
self.task.log_info(
"Going to prepare datasets based on the pre-coded regexps and annotations"
"entered by users...")
document_class = kwargs.get('document_class')
document_ids = kwargs.get('document_ids')
task_count = 0
if document_class:
task_count += BuildFieldDetectorDataset.build_sentences_to_fields_relations_dataset(
document_class, document_ids, self.task)
else:
for document_class, fields in DOCUMENT_FIELDS.items():
task_count += BuildFieldDetectorDataset.build_sentences_to_fields_relations_dataset(
document_class, document_ids, self.task)
self.task.update_subtasks_total(task_count)
