WinterMilestone 3 vaastav ReputationIdea: Suggest tasks based on history of completed tasks

From crowdresearch
Revision as of 17:05, 31 January 2016 by Vaastavanand (Talk | contribs)

(diff) ← Older revision | Latest revision (diff) | Newer revision → (diff)
Jump to: navigation, search

This page provides the details about the Reputation Idea I came up with for Winter Milestone 3


My idea is to suggest to workers tasks based on the profile or trend of tasks they generally tend to complete. This would basically looking at different types of data like the average time of completion, the kind of tasks a worker generally tends to perform, the average compensation he is looking from the task and the requester the worker generally tends to perform tasks for.

What I am basically suggesting is a task recommender system which finds optimal tasks for the worker based on the worker's task history.

How does it affect reputation

Benefits for the worker

I feel that it addresses workers' concerns regarding reputation in 2 key ways:

  • It saves worker the time and effort to find out tasks that they'd be interested in doing. A key observation from last week regarding the discussion about workers' needs was that they really cant afford to risk reputation. Thus they spend a lot of time finding the right tasks to complete so that they won't sacrifice their reputation. This recommender system would save the workers' their time which they can further spend on actually completing the tasks.
  • It provides workers with tasks that they would genuinely find interesting. Its a very simple psychological observation that people tend to perform better in things they are actually interested in. If we can provide workers with the tasks that they would find interesting then it would contribute to enhancing workers' reputation.

Benefits for the requester

In addition to addressing workers' concerns regarding reputation, this idea will also

  • The tasks posted by the requester are first being shown to the people who would actually be interested in doing the task. This would result in an increased quality of the results that the requester receives.
  • This also decreases the fake and cheat submissions people would receive. This is of course speculative but theoretically this seems like the more logical solution.


For old users, we could look at the last 50 tasks and learn about the kind of tasks they tend to work on.

For new users, we could ask them to fill a profile or form which will give us insight on what kind of tasks do they actually want to work on.