def upload_questions_to_crowdflower(title,
job_distinguish_tag,
question_content,
max_judgments_per_worker=50,
units_per_assignment=3,
judgments_per_unit=1,
payment_cents=3):
conn = crowdflower.Connection(api_key=api_key.api_key_string)
job = conn.upload(question_content)
job.update({
'title': title,
'max_judgments_per_worker': max_judgments_per_worker,
'units_per_assignment': units_per_assignment,
'judgments_per_unit': judgments_per_unit,
'payment_cents': payment_cents,
'instructions': '''
<h3><em><strong>Semantic Discrimination</strong> </em></h3>
<p><em><strong>Judge whether the two paragraphs describe the same object or not.</strong></em></p>
''',
'cml': '''
<p>{{content}}</p>
<cml:radios label="Are these two paragraphs describe the same object?" name="option" validates="required" gold="true">
<cml:radio label="Yes, they describe the same object" value="same"></cml:radio>
<cml:radio label="No, they describe the different object" value="different"></cml:radio>
</cml:radios>
''',
'options': {
'front_load': 0, # quiz mode = 1; turn off with 0
}
})
job.tags = [job_distinguish_tag]
job.launch(len(question_content), channels=('on_demand', 'cf_internal'))
def collect_answers_from_crowdflower(job_distinguish_tag):
conn = crowdflower.Connection(api_key=api_key.api_key_string)
result = []
out_dir = os.path.dirname(
os.path.realpath(__file__)) + os.path.sep + 'cf_temp/'
for job in conn.jobs():
# if job_distinguish_tag in job.tags and job.properties['state'] == 'finished':
if job_distinguish_tag in job.tags:
job.download_csv(out_dir + str(job.id) + '.csv')
reader = csv.reader(file(out_dir + str(job.id) + '.csv', 'rb'))
first_line = True
for line in reader:
if first_line == True:
first_line = False
for index, item in enumerate(line):
if item == 'id':
id_index = index
if item == 'category':
category_index = index
else:
worker_id = line[7]
unique_id = line[id_index]
option_single_answer = line[category_index].split('_')[1]
result.append((unique_id, worker_id, option_single_answer))
break
return job_distinguish_tag, result
def collect_answers_from_crowdflower(job_distinguish_tag):
conn = crowdflower.Connection(api_key=api_key.api_key_string)
rating_result = []
out_dir = os.path.dirname(
os.path.realpath(__file__)) + os.path.sep + 'cf_temp/'
for job in conn.jobs():
if job_distinguish_tag in job.tags:
job.download_csv(out_dir + str(job.id) + '.csv')
reader = csv.reader(file(out_dir + str(job.id) + '.csv', 'rb'))
first_line = True
for line in reader:
for index, item in enumerate(line):
print index, item
if item == 'id':
id_index = index
if item == 'option':
option_index = index
task_id = line[id_index]
option = line[option_index]
if option == 'same':
option_result = 1
elif option == 'different':
option_result = 0
worker_id = line[7]
if first_line == False:
rating_result.append([task_id, worker_id, option_result])
print[task_id, worker_id, option_result]
else:
first_line = False
return job_distinguish_tag, rating_result
def query_the_status_of_the_job(job_distinguish_tag):
conn = crowdflower.Connection(api_key=api_key.api_key_string)
for job in conn.jobs():
if job_distinguish_tag in job.tags:
status_of_the_job = job.properties['state']
break
print status_of_the_job
return status_of_the_job == 'finished'
def upload_questions_to_crowdflower(title,
job_distinguish_tag,
question_content,
max_judgments_per_worker=50,
units_per_assignment=3,
judgments_per_unit=1,
payment_cents=3):
conn = crowdflower.Connection(api_key=api_key.api_key_string)
refined_question_content = []
columns_content = question_content[0]['columns']
for i in range(len(question_content)):
new_dict = {}
for key in question_content[i]:
if key != 'columns':
new_dict[key] = question_content[i][key]
count = 0
for item in columns_content:
new_dict[str(item) + "answer_attr" + str(count)] = item
count += 1
refined_question_content.append(new_dict)
job = conn.upload(refined_question_content)
first_line = "<p>{{content}}</p>"
question_strings = first_line
for i in range(len(columns_content)):
item_first_line = "<p>{{" + columns_content[i] + "answer_attr" + str(
i) + "}}</p>"
item_second_line = '''<cml:text label="" validates="required" gold="true" name = "fill_''' + str(
i) + '''1" />'''
item = item_first_line + item_second_line
question_strings = question_strings + item
job.update({
'title': title,
'max_judgments_per_worker': max_judgments_per_worker,
'units_per_assignment': units_per_assignment,
'judgments_per_unit': judgments_per_unit,
'payment_cents': payment_cents,
'instructions': '''
<h3><em><strong>Semantic Discrimination</strong> </em></h3>
<p><em><strong>Judge whether the two paragraphs describe the same object or not.</strong></em></p>
''',
'cml': question_strings,
'options': {
'front_load': 0, # quiz mode = 1; turn off with 0
}
})
job.tags = [job_distinguish_tag]
job.launch(len(refined_question_content),
channels=('on_demand', 'cf_internal'))
def collect_answers_from_crowdflower(job_distinguish_tag):
conn = crowdflower.Connection(api_key=api_key.api_key_string)
result = []
out_dir = os.path.dirname(os.path.realpath(__file__)) + '/cf_temp'
for job in conn.jobs():
if job_distinguish_tag in job.tags:
job.download_csv(out_dir + os.path.sep + str(job.id) + '.csv')
reader = csv.reader(
file(out_dir + os.path.sep + str(job.id) + '.csv', 'rb'))
first_line = True
for line in reader:
if first_line == True:
for index, item in enumerate(line):
if item == 'id':
id_index = index
col_list = {}
for index, item in enumerate(line):
if "answer_attr" in item:
attr_string = item.split("answer_attr")[1]
if '_dot_' in item:
attr_col_string_list = item.split(
"answer_attr")[0].split('_dot_')
attr_col_string = attr_col_string_list[
0] + '.' + attr_col_string_list[1]
col_list[int(attr_string)] = attr_col_string
elif '_dot_' not in item:
col_list[int(attr_string)] = item.split(
"answer_attr")[0]
fill_list = []
for index, item in enumerate(line):
if 'fill' in item and 'gold' not in item:
number_this_time = int(item[-2])
fill_list.append((number_this_time, index))
first_line = False
else:
worker_id = line[7]
attr_answers = {}
unique_id = line[id_index]
for i in range(len(fill_list)):
attr_answers[col_list[i]] = line[fill_list[i][1]]
if first_line == False:
result.append((unique_id, worker_id, attr_answers))
else:
first_line = False
return job_distinguish_tag, result
def upload_questions_to_crowdflower(title,
job_distinguish_tag,
question_content,
max_judgments_per_worker=50,
units_per_assignment=3,
judgments_per_unit=1,
payment_cents=3):
conn = crowdflower.Connection(api_key=api_key.api_key_string)
job = conn.upload(question_content)
options = question_content[0]['options']
first_line = '''
<img src="{{url}}" id="">
<p>{{content}}</p>
<cml:radios validates="required" label="Choose the best option" name="category" class="">
'''
question_strings = first_line
for i in range(len(options)):
item = '''<cml:radio label="''' + str(
options[i]) + '''" value = "fill_''' + str(
options[i]) + '''" />'''
question_strings = question_strings + item + '\n'
question_strings = question_strings + '''</cml:radios>'''
job.update({
'title': title,
'max_judgments_per_worker': max_judgments_per_worker,
'units_per_assignment': units_per_assignment,
'judgments_per_unit': judgments_per_unit,
'payment_cents': payment_cents,
'instructions': '''
<h3><em><strong>Tell us the details about the country</strong> </em></h3>
''',
'cml': question_strings,
'options': {
'front_load': 0, # quiz mode = 1; turn off with 0
}
})
job.tags = [job_distinguish_tag]
job.launch(len(question_content), channels=('on_demand', 'cf_internal'))
def upload_questions_to_crowdflower(title, job_distinguish_tag,
question_content, max_judgments_per_worker,
units_per_assignment, judgments_per_unit,
payment_cents):
conn = crowdflower.Connection(api_key=api_key.api_key_string)
refined_question_content = []
columns_content = question_content[0]['columns']
for i in range(len(question_content)):
limit = int(question_content[i]['limit'])
repeats = int(question_content[i]['repeats'])
total_repeat_times = limit * repeats
new_dict = {}
for key in question_content[i]:
if key != 'columns':
new_dict[key] = question_content[i][key]
count = 0
for item in columns_content:
if '.' in item:
first_part = item.split('.')[0]
second_part = item.split('.')[1]
new_dict[str(first_part) + '_dot_' + str(second_part) +
'answer_attr' + str(count)] = item.split('.')[1]
count += 1
elif '.' not in item:
new_dict[str(item) + 'answer_attr' + str(count)] = item
count += 1
new_dict['number_counting'] = str(i)
for j in range(total_repeat_times):
refined_question_content.append(new_dict)
job = conn.upload(refined_question_content)
first_line = "<p>{{content}}</p>"
question_strings = first_line
for i in range(len(columns_content)):
if '.' in columns_content[i]:
first_part = str(columns_content[i].split('.')[0])
second_part = str(columns_content[i].split('.')[1])
item_first_line = "<p>{{" + first_part + '_dot_' + second_part + 'answer_attr' + str(
i) + "}}</p>"
item_second_line = '''<cml:text label="" validates="required" gold="true" name = "fill_''' + str(
i) + '''1" />'''
item = item_first_line + item_second_line
question_strings = question_strings + item
elif '.' not in columns_content[i]:
item_first_line = "<p>{{" + columns_content[
i] + "answer_attr" + str(i) + "}}</p>"
item_second_line = '''<cml:text label="" validates="required" gold="true" name = "fill_''' + str(
i) + '''1" />'''
item = item_first_line + item_second_line
question_strings = question_strings + item
job.update({
'title': title,
'max_judgments_per_worker': max_judgments_per_worker,
'units_per_assignment': units_per_assignment,
'judgments_per_unit': judgments_per_unit,
'payment_cents': payment_cents,
'instructions': '''
<h3><em><strong>Tell us the details about the country</strong> </em></h3>
''',
'cml': question_strings,
'options': {
'front_load': 0, # quiz mode = 1; turn off with 0
}
})
job.tags = [job_distinguish_tag]
job.launch(len(refined_question_content),
channels=('on_demand', 'cf_internal'))
import os
import sys
import json
import random
import crowdflower
from crowdflower.exception import CrowdFlowerError
# expects api key to be available in your environment variables; does not use cache
conn = crowdflower.Connection()
job_tag = 'python-example'
def _find_job():
for job in conn.jobs():
if job_tag in job.tags:
return job
def create():
filename = os.path.join(crowdflower.root, 'examples', 'spam.txt')
def iter_data(labels=None):
for i, line in enumerate(open(filename), 1):
label, text = line.strip().split('\t', 1)
if labels is None or label in labels:
# the "text" key is the only required field; the others just help us
# keep track of what's been annotated and mark which are the test data
yield {
'id': '{}:{}'.format(filename, i),
'text': text,
'label': label
