Task Classification: Towards Tasks Correlation for Combinatorial Auction Mechanism in Crowdsourcing

Abstract

People experience a revolution of society when the crowdsourcing period comes. The social reform changed the way of work and lifestyle. And the popularity of crowdsourcing has brought a new opportunity for mobile computing and online auctions. Because of the significance of incentivizing worker participation in crowdsourcing system, the correlative theories of auction-based incentive mechanisms have been proposed in past literature. Combinatorial auctions can be used to reach efficient resource and task allocations in Multi-agent systems where the items are complementary or substitutable. Although many good auction-based incentive mechanisms can attract more participation, facing a large-scale of complex social perception tasks in the market, a combinatorial auction mechanism which simply relies on to allocate resources is not efficiently. In order to solve this problem, we design two algorithm models to classify all crowdsourcing-tasks into many groups according to the intentions that the workers provided, and then the workers and the crowdsourcing platform start the combinatorial auction for each task-group. The first model, namely the frequent-driven model, adopts correlation analysis technology, and the second model, namely weight-driven model, use a method that combines the concept of weight and the hierarchical clustering technology to solve problems.