Crowdsourced Intelligence

• 

  Confluence Page with Crowdsourced Intelligence Plugin

• 

  Mechanical Turk Task Created by Crowdsourced Intelligence Plugin

Inspiration

Crowdsourced Intelligence was born to enable gathering or crowdsourcing intelligence in a quick and automated way for tasks. As a PM, when I’m managing a feature or a project, sometimes there are tasks that need to be completed by a third party that requires human input in order to move that feature along. However, it’s cumbersome to have to manage work outside of your workflow with other tools. That’s how Crowdsourced Intelligence plugin was born. It essentially allows you to pull in a workflow that involves gathering human intelligence tasks right into your existing workflow.

What it does

The plugin allows you to create a task that gets posted to Amazon's Mechanical Turk -- a human intelligence platform for performing manual tasks -- and feeds back in your requested data (i.e. intelligence). In the plugin, you can set specific parameters such as budget, number of responses and the request itself. Rather than having to go onto another platform, in this case Mechanical Turk, this brings that workflow directly into Confluence to make it easier to have one central place to gather and store information. It also allows for easily transferring that same data to developers later on in Jira since all the required info is in one place.

For the demo, I specifically focused on using this plugin for translations. However, because users have the ability to create their own template, this plugin is extensible and customizable to almost any use case that requires gathering (crowdsourcing) data (intelligence) from a third party platform such as Mechanical Turk.

How I built it

Used forge to build the component on React and deploy it on Confluence. A backend server was the used to sync the tasks and uploads and the jobs.

Challenges I ran into

Learning React and how that plays in with the new Forge framework. Fortunately documentation was a breeze and it was fairly easy to make it work.

What I learned

React components is a very powerful base in which to extend from and it was cool to build something to put this to a practical application.

What's next for Crowdsourced Intelligence

• Create more use cases through templates
• More parameters for customizing tasks
• Have an automatic notification sent when all your required responses have been completed
• Make the backend component serverless and just deploy it without server (ie. via Amazon or other platform)