def run(self, execution_context):
filtered_word_types = self.get_config('train.filtered_word_types')
data_name = execution_context.event.get('data_name', 'default')
text = execution_context.event.get('text')
if text is None:
raise ValueError('text cannot be null')
config = persist.get_data_config(data_name)
tokenizer = TokenizerFactory.get_tokenizer(config.get('tokenizer'))
# clean text if requested
if config.get('clean_text'):
text = pre_process.clean_text(text, filtered_word_types)
# tokenize words and predict
result_word, result_proba = classifier.predict(text, data_name)
result_proba *= 100
# pos tagging the text
tagged_text = self.tag_text(tokenizer(text))
_, content = self.send_msg(result_word, result_proba, tagged_text)
content_obj = json.loads(content.decode('utf-8'))
result = 'predict: ' + result_word + ' with probability: ' \
+ str((int)(result_proba)) + '%. response from bot: ' \
+ str(content_obj.get('msg')) \
+ ' [tagged words: ' + ' ' \
+ ' '.join(list(map(lambda w: '(' + w[0] + ' ' + w[1] + ')', tagged_text))) \
+ ']'
execution_context.finish({
'raw': result,
'bot_response': content_obj.get('msg')
})
def run(execution_context):
"""run the action"""
filtered_word_type = app.get_config('predict.filtered_word_types')
remove_stop_words = app.get_config('train.remove_stop_words')
data_name = execution_context.event.get('data_name', 'default')
text = execution_context.event.get('text')
if text is None:
raise ValueError('text cannot be null')
config = persist.get_data_config(data_name)
tokenizer = TokenizerFactory.get_tokenizer(config.get('tokenizer'))
if config.get('clean_text'):
text = pre_process.clean_text(text, remove_stop_words)
tokenized_text = tokenizer(text)
result_word, result_proba = pre_process.predict(tokenized_text, data_name)
LOGGER.warning('predict %s with probability %2f %%', result_word,
result_proba * 100)
pos_tagged_text = nltk.pos_tag(nltk.word_tokenize(text))
filtered_text = \
[(w, word_type) for w, word_type in pos_tagged_text if not word_type in filtered_word_type]
send_msg(result_word, result_proba, filtered_text)
def run(self, execution_context):
data_name = execution_context.event.get('data_name', 'default')
config = persist.get_data_config(data_name)
reader = self.open_test_file(data_name)
result = self.test_data(reader, data_name, config)
execution_context.finish(result)
def run(execution_context):
"""run the action"""
data_name = execution_context.event.get('data_name', 'default')
config = persist.get_data_config(data_name)
with open('cache/data/' + data_name + '/raw.csv') as data_file:
reader = csv.reader(data_file)
train_data(reader, data_name, config)
def run(self, execution_context, app_context):
"""run the action"""
data_name = execution_context.event.get('data_name', 'default')
config = persist.get_data_config(data_name)
with open('cache/data/' + data_name + '/raw.csv') as data_file:
reader = csv.reader(data_file)
self.train_data(reader, data_name, config, app_context)
execution_context.finish('train done')
def run(self, execution_context, app_context):
"""run the action"""
data_name = execution_context.event.get('data_name', 'default')
config = persist.get_data_config(data_name)
filtered_word_types = app_context.get_config(
'train.filtered_word_types')
with open('cache/data/' + data_name + '/test.csv') as data_file:
reader = csv.reader(data_file)
result = self.test_data(reader, data_name, config,
filtered_word_types)
execution_context.finish(result)
def run(self, execution_context, app_context):
"""run the action"""
event = execution_context.event
data_name = event.get('data_name', 'default')
text = event.get('text')
self.validate(text, app_context)
data_config = persist.get_data_config(data_name)
if not data_config.get('allow_teaching'):
raise ValueError('data_name ' + data_name +
' is not allowed to teach')
with open("cache/data/" + data_name + '/raw.csv', 'a') as data_file:
data_file.write(text + '\n')
with open("cache/data/" + data_name + '/test.csv', 'a') as data_file:
data_file.write(text + '\n')
execution_context.finish('teach done')