Journal of Northeastern University Natural Science ›› 2014, Vol. 35 ›› Issue (8): 1097-1101.DOI: 10.12068/j.issn.1005-3026.2014.08.008

• Information & Control • Previous Articles     Next Articles

Crowdsourcing Quality Evaluation Strategies Based on Voting Consistency 〓

YUE Dejun1, YU Ge1, SHEN Derong1, YU Xiaocong2   

  1. 1 School of Information Science & Engineering, Northeastern University, Shenyang 110819, China; 2 Department of Computer Crime Detection, China Criminal Police College, Shenyang 110035, China.
  • Received:2013-10-30 Revised:2013-10-30 Online:2014-08-15 Published:2014-04-11
  • Contact: YUE Dejun
  • About author:-
  • Supported by:
    -

Abstract: A dynamic evaluation strategy for the worker precision based on the voting consistency is proposed. The crowdsourcing task set is divided into multiple task subsets and each of them is regarded as a task phase. When a task phase is finished, the whole votes and the number of the worker’s voting value agreeing with the working group’s one are recorded. Then, the workers’precision is evaluated and the unqualified workers are detected. All the unqualified workers are dynamically replaced to ensure that the correct rate of the group exceeds the task accuracy threshold. The effectiveness of the proposed method is evaluated using three realworld datasets, which shows that the method can achieve good performance.

Key words: crowdsourcing, voting, consistency, quality evaluation, dynamical replacement

CLC Number: