Milestone 3 pentagram DarkHorseIdea: Task Recommender System

From crowdresearch
Jump to: navigation, search
  • Integrate a recommender system with the crowdsourcing platform.
  • Typically each worker has a list of recommended tasks based on skills, current work available, reputations etc.
  • The recommender system will also consider the previous tasks completed by workers and then accordingly give corresponding recommendations
  • This will reduce the time a worker spends on finding new tasks (one of the common complaints).
  • This can be implemented for the requesters also, a list of workers are recommended (might be an unfair method)

A proposed algorithm

Darkhorse recommender system.png

Description (Contest-based recommendation algorithm)

  • Recommendations for Worker 5 is first identified by finding the task done by him (Task 2)
  • A task similar to Task 2 is Task 3 (Category 2)
  • Now, Task 3 is also completed by Worker 3 and Worker 4
  • We find task done in common by Worker 3 & 4 i.e. Task 5
  • Thus Task 5 is recommended to Worker 5