Today, i am going to write about my first experience with Amazon Mechanical Turk.
Question? What is Amazon Mechanical Turk?
It’s a crowdsourcing internet marketplace that enables programmers like us to use the human intelligence to perform tasks that softwares or computer can not perform. It is widely known as MTurk. It has two main entities, Requester and Worker. For instance we need to tag 1000 photos with appropriate tags, this is something a computer or software can’t do by itself unless there is some heavy use of Augmented Reality stuff is involved. In the given scenario what we can do is just create a task on MTurk known as HIT (Human Intelligence Task) and let other people (crowd) to do the hard part. Requester would be the one who’ll responsible for creating the HIT(Human Intelligence Task) and Worker would be the one who’ll tag the photos. In exchange requester will pay some amount of money to workers.
To read more about it follow the link: – Amazon Mechanical Turk
So for one of my projects i had to embed this technology into my app and i would love to share my first encounter with it.
Day 1: – GOOD NEWS!! Found they have SDKs. I was so happy to see that as i thought it is going to remove the complexity.
Day 2: – Trying to figure out how to use the SDK. I ran few of the samples provided with SDK.
Day 3: – SDK turned into a dark tunnel with no end due to some compatibility issues with other technologies i was using in my app.
Day 4: – Tried to modify the SDK to make it compatible with other technologies.
Day 5: – Gone mad modifying it. Error one after another.
Day 6: – Saw a light in tunnel called REST APIs directly. I followed that light and lived happily after (So far :S).
So finally on sixth day i was able to make things work. In my next blog, i am going to put a small tutorial on how to use MTurk REST APIs.