An Explorative Approach for Crowdsourcing Tasks Design

摘要: Crowdsourcing applications are becoming widespread; they cover very different scenarios, including opinion mining, multimedia data annotation, localised information gathering, marketing campaigns, expert response and so on. The quality of the outcome these depends on design parameters constraints, it is hard to judge about their combined effects without doing some experiments; other hand, there no experiences or guidelines that tell how conduct experiments, thus often conducted in an ad-hoc manner, typically through adjustments initial strategy may converge a parameter setting which quite from best possible one. In this paper we propose comparative, explorative approach for designing crowdsourcing tasks. method consists defining representative set execution strategies, then execute them small dataset, collect measures each candidate strategy, finally decide be used with complete dataset.