CommunitySourcing

摘要: Online labor markets, such as Amazon's Mechanical Turk, have been used to crowdsource simple, short tasks like image labeling and transcription. However, expert knowledge is often lacking in making it impossible complete certain classes of tasks. In this work we introduce an alternative mechanism for crowdsourcing that require specialized or skill: communitysourcing --- the use physical kiosks elicit from specific populations. We investigate potential by designing, implementing evaluating Umati: vending machine. Umati allows users earn credits performing using a touchscreen attached Physical rewards (in case, snacks) are dispensed through traditional mechanics. evaluated whether can accomplish grade Computer Science exams. placed university building, targeting students with grading snacks. Over one week, 328 unique (302 whom were students) completed 7771 (7240 students). 80% had never participated market before. found was able exams 2% higher accuracy (at same price) at 33% lower cost equivalent accuracy) than single-expert grading. Turk workers no success These results indicate successfully high-quality communities.