This tool permits the creation of multiple Tasks of multiple HITs (images or videos) in Amazon Mechanical Turk.
- Sign in as a Requester here and create an Amazon account.
- Create New Project here selecting Other type for the template.
- Follow this image:
- Copy&Paste the content of layoutImages.html in DESIGN LAYOUT (pass 2 for the creation).
- Save.
- Copy HITTypeID and LayoutID from:
- Paste HITTypeID and LayoutID in Key.py.
Repeat 2...7 with layoutVideo.html.
- Create AWS account here.
- Set up IAM console here.
- Follow this and click on ADD USER:
- Enter text like and press Next button.
- Select AmazonMechanicalTurkFullAccess like:
- Insert new key like:
- Select Next, and Create User.
- Copy the two keys and paste it in Key.py:
- Sign in here with AWS UserID and password created earlier.
- Create (in different times) two new buckets named:
imagesformturk
andvideosformturk
. - Deselect
Block all public access
here: - Press Create Bucket.
- In
imagesformturk
open bucket -> go to authorization -> Policy Bucket and paste:
"Version": "2012-10-17",
"Statement": [
{
"Sid": "PublicReadGetObject",
"Effect": "Allow",
"Principal": "*",
"Action": "s3:GetObject",
"Resource": "arn:aws:s3:::imagesformturk/*"
}
]
}
- In
videosformturk
open bucket -> go to authorization -> Policy Bucket and paste:
{
"Version": "2012-10-17",
"Statement": [
{
"Sid": "PublicReadGetObject",
"Effect": "Allow",
"Principal": "*",
"Action": "s3:GetObject",
"Resource": "arn:aws:s3:::videosformturk/*"
}
]
}
- Download Xampp here
- After having installed go to
http://localhost/phpmyadmin/
and create DB with these specifics: User:"utente" Password: "pass123" DB Name:"dbmysql"
- Uncomment in createImmHIT.py, resultsImm.py, createVideoHIT.py, resultsVid.py the line to connect to the live marketplace instead of the sandbox.
Now it's all setted up! The tool can work correctly. Use the requirement.txt to recreate the virtual env for the project.
- Create a new Task, choosing from Image or Video Tasks.
- Select files to upload on Amazon MTurk.
- View of all Tasks precedently created.
- It's possible to refresh the results of workers.
- Each Task has a personalized dashboard with adaptive queries that shows the results.
- It's possible to create other Tasks from the home view.
- Evaluation History of single file (of the selected Task). Type: Line Graph.
- Evaluation History of single worker (that evaluates some HITs in the task selected). Type: Line Graph.
- Comparison of the evaluation done by two different workers (that evaluates some HITs in the task selected). Type: Radar Graph.
- Single file evaluation results. Type: Pie Graph.
- Workers ordered by the number of submitted HITs. Type: Bar Graph.
- Number of submitted HITs by workers and average of the evaluation vote that they have assigned. Type: Scatter Graph.
- Liar workers ordered by the number of times that they assigned random value in age and sex field. Type: Bar Graph.
- Results showed based on sex: male/female. Type: Pie Graph.
- Percents of the screen resolution of the workers. Type: Bar Graph.
- Percents of the age of the workers. Type: Bar Graph.
- Average and standard deviation of each file evaluated in the task; and average of all the votes of the task. Type: Line Graph.
- Modular multiple graph with 3 of the precedent views.
Edoardo Re, 2019