A Survey on Task Assignment in Crowdsourcing

PhD Qualifying Examination


Title: "A Survey on Task Assignment in Crowdsourcing"

by

Mr. Yanrong KANG


Abstract:

Crowdsourcing systems have emerged as an effective paradigm of harnessing human 
computing power to solve problems which are difficult for computers. Its 
popularity and promise has attracted great attention from researcher 
communities to optimize various aspect of crowdsourcing markets. In 
crowdsourcing paradigm, tasks are distributed to a large group of online 
workers. However, due to human error and the prevalence of spam, answers 
collected from crowds are often noisy. Redundancy is used to control the 
quality of completed tasks, by assigning each task to multiple workers and 
combining their answers. Here arises the task assignment problem, which aims to 
obtain optimal tradeoff between quality and redundancy.

In this paper, we survey the state-of-the-art approaches on task assignment in 
crowdsourcing. These approaches formulate the problem under different settings 
and according to the way how they utilize worker evaluation, they are 
classified to three categories: assignment before evaluation, evaluation during 
assignment and assignment after evaluation.
  To introduce problem background, the basic architecture of crowdsourcing 
systems and the quality control problem are presented first. Then different 
problem settings are introduced. After surveying methods in three categories, I 
end this survey by identifying some possible steps forward.


Date:			Wednesday, 26 March 2014

Time:                   4:00pm - 6:00pm

Venue:                  Room 3501
                         Lifts 25/26

Committee Members:	Dr. Lei Chen (Supervisor)
 			Dr. Yunhao Liu (Supervisor)
  			Dr. Lin Gu (Chairperson)
  			Dr. Pan Hui
  			Dr. Ke Yi


**** ALL are Welcome ****