turkleton 0.1.1

Features

• Interfaces for uploading assignments from pre-built layouts.

• ORM-like interface for downloading and evaluating assignment results.

Examples

Here’s how you would define a task template to be uploaded:

from turkleton.assignment import task

class MyTask(task.BaseTask):

    __layout_id__ = 'MY LAYOUT ID'
    __reward__ = 0.25
    __title__ = 'Guess How Old From Picture'
    __description__ = 'Look at a picture and guess how old the person is.'

Now that we’ve defined our assignment we can easily upload it to Mechanical Turk:

from turkleton import connection

conn = connection.make_connection(AWS_ACCESS_KEY, AWS_SECRET_ACCESS_KEY)
task = MyTask({'param1': 'val1', 'param2': 'val2'})
hit = task.upload(conn, batch_id='1234')

This will create a new task and upload it to Mechanical Turk. The optional batch_id parameter allows you to set the annotation for the task to an arbitrary string that you can use to retrieve tasks later in batches.

History

0.1.0 (2015-01-11)

• First release on PyPI.