Crowdsourced Databases: Query Processing with People

摘要: Amazon’s Mechanical Turk (\MTurk") service allows users to post short tasks (\HITs") that other can receive a small amount of money for completing. Common on the system include labelling collection images, combining two sets images identify people which appear in both, or extracting sentiment from corpus text snippets. Designing workow various kinds HITs ltering, aggregating, sorting, and joining data sources together is common, comes with set challenges optimizing cost per HIT, overall time task completion, accuracy MTurk results. We propose Qurk, novel query managing these workows, allowing crowdpowered processing relational databases. describe number execution optimization challenges, discuss some potential solutions.