Independent Worker Selection In Crowdsourcing

Ang Li, Wenjun Jiang, Xueqi Li, Xinrong Chen, Guojun Wang

Abstract: Crowdsourcing allows workers to participate in collaborative works. The selection of proper workers is essential for the quality of task completion. However, in crowdsourcing mode, task plagiarism or collusion among connecting workers greatly affects the quality of some specific tasks. In this paper, we propose an independent worker selection framework based on the technology of community detection and link prediction. The former is used to guarantee that the selected workers have no or very weak connections currently, the latter is to guarantee that they will have no connections in future. We also conduct intensive homogeneity analysis to help guide the independence among the selected workers. Experimental results validate that our work can select workers with high efficiency and high independence.