def train_model_from_es_data(self):
"""
This method is used to train the crf model by first extracting training data from ES
for the entity and training the crf model for the same.
"""
datastore_object = DataStore()
ner_logger.debug('Fetch of data from ES for ENTITY: %s started' %
self.entity_name)
result = datastore_object.get_crf_data_for_entity_name(
entity_name=self.entity_name)
sentence_list = result.get(SENTENCE_LIST, [])
entity_list = result.get(ENTITY_LIST, [])
if not sentence_list:
raise ESCrfTrainingTextListNotFoundException()
if not entity_list:
raise ESCrfTrainingEntityListNotFoundException()
ner_logger.debug('Fetch of data from ES for ENTITY: %s completed' %
self.entity_name)
ner_logger.debug('Length of text_list %s' % str(len(sentence_list)))
model_path = self.train_crf_model_from_list(
entity_list=entity_list, sentence_list=sentence_list)
return model_path
def get_crf_training_data(request):
"""
This function is used obtain the training data given the entity_name.
Args:
request (HttpResponse): HTTP response from url
Returns:
HttpResponse : With data consisting of a dictionary consisting of sentence_list and entity_list
Examples:
get request params
key: "entity_name"
value: "city"
"""
response = {"success": False, "error": "", "result": []}
try:
entity_name = request.GET.get(ENTITY_NAME)
datastore_obj = DataStore()
result = datastore_obj.get_crf_data_for_entity_name(entity_name=entity_name)
response['result'] = result
response['success'] = True
except (DataStoreSettingsImproperlyConfiguredException,
EngineNotImplementedException,
EngineConnectionException, FetchIndexForAliasException) as error_message:
response['error'] = str(error_message)
ner_logger.exception('Error: %s' % error_message)
return HttpResponse(json.dumps(response), content_type='application/json', status=500)
except BaseException as e:
response['error'] = str(e)
ner_logger.exception('Error: %s' % e)
return HttpResponse(json.dumps(response), content_type='application/json', status=500)
return HttpResponse(json.dumps(response), content_type='application/json', status=200)