CDAS is a Crowdsourcing Data Analytics System. The objectives are to design and implement an effective system to exploit the crowd intelligence for improving the performance of different data analytics jobs, such as web table integration, image tagging, sentiment analysis, etc.
The system consists of three layers: Platform Layer, Service Layer and Application Layer, which are shown below:
The Platform Layer provides basic funtions to facilitate effective interaction between requesters and workers. Specifically, requesters can publish crowdsourcing tasks on the platform using Web interface or CDAS API functions. On the other hand, workers can find and complete tasks to earn money. The Platform Layer also provides APIs to support the Service Layer and Application Layer, which are explained below.
1 The Service Layer is designed to provide high quality data to the application layer. Specifically, the Service Layer aims to provide high accurate results while keeping the cost of the platform low. It also sets the proper price and selects the best workers for the crowdsourcing tasks in order to let the customer benefit more from the crowdsourcing system.
2 The Application Layer is designed for effective collaboration between Crowd and Machine. It provides a set of semantic operators to help requesters decompose high-level jobs into suitable microtasks for the crowd. Meanwhile, it selects the microtasks to be published for crowdsourcing and designs active learning algorithms for accurate inference.