Getting More for Less: Optimized Crowdsourcing with Dynamic Tasks and Goals

摘要: In crowdsourcing systems, the interests of contributing participants and system stakeholders are often not fully aligned. Participants seek to learn, be entertained, perform easy tasks, which offer them instant gratification; want users complete more difficult bring higher value crowdsourced application. We directly address this problem by presenting techniques that optimize process jointly maximizing user longevity in true derives from participation. first present models predict "survival probability" a at any given moment, is, probability will proceed next task offered system. then leverage survival model dynamically decide what assign motivating goals user. This allows us for short term (getting tasks done) long (keeping engaged longer periods time). show assigning significantly increases an extensive empirical evaluation, we observed our allocation strategy amount information collected up 117.8%. also explore utility with goals. demonstrate setting specific, static can highly detrimental long-term participation, as completion goal (e.g., earning badge) is common drop-off point many users. dynamically, conjunction judicious 249%, compared existing baselines use fixed objectives.