Published:
2013-11-10
Proceedings:
Proceedings of the AAAI Conference on Human Computation and Crowdsourcing, 1
Volume
Issue:
Vol. 1 (2013): First AAAI Conference on Human Computation and Crowdsourcing
Track:
Works in Progress
Downloads:
Abstract:
We present our ongoing research on improving the task presentation using cognitively inspired features to optimize the performance of crowd workers. For the task of extracting text from scanned images, we generated three task-presentation designs by modifying two features of the task - visual saliency of target fields and working memory requirements. Our experiments conducted on Amazon Mechanical Turk (AMT) indicate that modifying visual saliency results in better performance.
DOI:
10.1609/hcomp.v1i1.13103
HCOMP
Vol. 1 (2013): First AAAI Conference on Human Computation and Crowdsourcing
ISBN 978-1-57735-607-3