International Journal of Computer Applications |
Foundation of Computer Science (FCS), NY, USA |
Volume 52 - Number 4 |
Year of Publication: 2012 |
Authors: Hanieh Javadi Khasraghi, Shahriar Mohammadi |
10.5120/8190-1556 |
Hanieh Javadi Khasraghi, Shahriar Mohammadi . An Innovative Crowdsourcing Approach for Amazon Mechanical Turk. International Journal of Computer Applications. 52, 4 ( August 2012), 20-25. DOI=10.5120/8190-1556
Web2 and the evolving vision of Web3 have a great effect on facilitation of information sharing, information aggregation, interoperability, user-centered design, collaboration on the World Wide Web, and crowd-centered services. New concept of Web is the intuition that drives crowdsourcing, crowd servicing, and crowd computing. With crowdsourcing emergence people get motivated to work through internet without being limited by time or geographical location. On the other hand employers could have their jobs done faster and cheaper. This paper is going to introduce an innovative approach for Amazon Mechanical Turk (AMT) crowdsourcing marketplace. In current AMT marketplace, workers especially new ones need to qualify themselves for each requester that has submitted Human Intelligence Tasks (HITs) in AMT, and there is lack of shared reputation system; some workers may cheat on tasks in order to maximize their income, as a result requesters are uncertain of the quality of results, so they offer lower rewards and consequently qualified workers leave the marketplace. Because of the above shortcomings, we introduce a new approach for AMT crowdsourcing marketplace. In our proposed approach we offer to distribute HITs among Amazon's customers and ask them to work on tasks in exchange for discount. The distribution of HITs is based on customers' interests and skills that Amazon has this information in its database. Through our proposed approach the HITs will be done by more qualified people, and spammers will be decreased to the minimum. This innovative approach is very efficient, time saving, and user friendly (because workers don't need to search for HITs of their interests).