Service Layer: Cost-Sensitive Quality Control
The service layer is designed to provide the cost-sensitive quality control in the CDAS system, i.e. returning high accuracy results while keeping the cost low. It also sets the proper price for the crowdsourcing tasks in order to let customers benefit more from
the crowdsourcing system.
The overview of the service layer is shown above, containing four components:
The cost controller guarantees that the quality of results is significantly improved while the processing cost is effectively reduced at the same time.
The quality controller considers the value of this question, the risk of getting incorrect answers and the cost of workers in the crowdsourcing system to
analyze and decide if a crowdsourcing task can be stopped before collecting all the answers in real-time.
The price policy model finds a proper price policy such that the cost is minimized while the quality of the results still satisfied and all HITs are finished in time.
The worker selector chooses the best workers based on their performance to improve the accuracy of the crowdsourcing results.