Task recommendation in crowdsourcing systems

• DOI: 10.1145/2442657.2442661
• Publication Date: 2013-02-22T19:25:33Z
• Authors: Man-Ching Yuen; Irwin King; Kwong-Sak Leung
• Keywords: Crowdsourcing; Knowledge worker; Task Analysis

In crowdsourcing systems, tasks are distributed to networked people complete such that a company's production cost can be greatly reduced. Obviously, it is not efficient the amount of time for worker spent on selecting task comparable with working task, but monetary reward just small amount. The available history makes possible mine workers' preference and provide favorite recommendations. Our exploratory study survey results collected from Amazon Mechanical Turk (MTurk) shows histories reflect preferences in systems. Task recommendation help workers find their right faster as well requesters receive good quality output quicker. However, previously proposed classification based approach only considers performance history, does explore searching history. our paper, we propose framework modeling preference-based recommendation, aiming recommend who likely prefer work accepted by requesters. We consider both more accurately. To best knowledge, first use matrix factorization