Personalized and Quality-Aware Task Recommendation in Collaborative Crowdsourcing

Kun Lu, Jiaxi Wang, Mingchu Li, Zhiheng Zhang

Abstract: Crowdsourcing is a promising solution of collaborative computing aiming at solving settling problems within distributed environment. A collaborative crowdsourcing system(CCS) usually consists of a platform, users performing tasks and thousands or even millions of tasks, each of which is usually simple such as choice making or item rating. However, existing works always match users and tasks without addressing users' interest, which could lead to a reduction of users' enthusiasm involved in the subsequent tasks. In addition, researchers often ignore the truth inference while completing task matching, which should be both significant in CCS. To this end, we jointly investigate the task matching and truth inference in CCS. We propose an integrated framework that can motivate users' participates while guaranteeing the quality of tasks. Finally, extensive simulations illustrate that our algorithm has an outstanding performance on both task matching and truth inferring.