WinterMilestone 4 freddiev Task Rank Proposal

From crowdresearch
Jump to: navigation, search

Task Rank Proposal

Currently on many crowdsourcing platforms, requesters and workers do not get matched up with their own ideal candidates; that is to say, requesters don't always get matched to a worker that will complete tasks with the quality that the requester has in mind, and the worker doesn't always get matched with a requester that will carefully lay out all of the tasks and say what is expected of the worker.

Requesters currently try to post content such as videos to explain what tasks need to be done, but the tasks don't have any labels attached to them that show difficulty, time to complete, etc etc. This hasn't worked because workers go in and accept tasks without knowing how everything is split up beforehand. As a result, when workers complete tasks, the requesters don't always get the work they wanted because they think the work hasn't been done properly, completely, with high quality, etc, and the workers have just wasted both the requesters time, and their own time.

In order to solve this sort of issue, the general idea is to have a set way of organizing tasks by difficulty, type of skills needed, estimated length to complete, etc. This is so that workers know what they're getting themselves into, and also so the workers have a better chance of having tasks completed by workers who meet expectations.

In a running system, this is how I envision it might work:


  • The requester lists out tasks individually, with individual videos for each task
 * The videos include: what the task is and what sort of expectations there are
  • The requester labels the tasks with categories
 * Examples:
    Computer Science: programming, design, UI/UX, machine learning, data mining
    Finance: analysis, algorithmic trading, investment portfolios
    Art: painting, drawing, graphic design
  • The requester also labels tasks with what sort of skills are needed (i.e., javascript, photoshop, html/css)
  • Through some sort of supervised algorithm or data mining, determine how long tasks should take based off of previously collected data and assign a difficult/estimated time category to the tasks listed out by the requester


  • Look through which tasks have been listed by requesters (that they are able to see)
  • Be able to pick and choose which tasks they can assign themselves to
  • Complete tasks based off the individual videos for that specific task

Matching workers & requesters

  • Based off Boomerang/worker ratings, allow the workers to see a certain number of tasks; from one difficulty level up to the highest one they'd be qualified for