MTurk Crowdsourcing: A Viable Method for Rapid Discovery of Arabic Nicknames?
Chiara Higgins, Elizabeth J. McGrath, Laila Moretto · 2010
This paper presents findings on using crowdsourcing via Amazon Mechanical Turk (MTurk) to obtain Arabic nicknames as a contribution to exiting Named Entity (NE) lexicons. It demonstrates a strategy for increasing MTurk participation from Arab countries. The researchers validate the nicknames using experts, MTurk workers, and Google search and then compare them against the Database of Arabic Names (DAN). Additionally, the experiment looks at the effect of pay rate on speed of nickname collection and documents an advertising effect where MTurk workers respond to existing work batches, called Human Intelligence Tasks (HITs), more quickly once similar higher paying HITs are posted. 1