Dropout Prediction in Crowdsourcing Markets
This addresses dropout prediction for crowdsourcing platforms, which is incremental as it builds on existing understanding of worker behavior.
The paper tackles the problem of predicting worker dropout in crowdsourcing markets, showing that it is possible to predict dropouts based on success rates and arrival patterns of workers.
Crowdsourcing environments have shown promise in solving diverse tasks in limited cost and time. This type of business model involves both the expert and non-expert workers. Interestingly, the success of such models depends on the volume of the total number of workers. But, the survival of the fittest controls the stability of these workers. Here, we show that the crowd workers who fail to win jobs successively loose interest and might dropout over time. Therefore, dropout prediction in such environments is a promising task. In this paper, we establish that it is possible to predict the dropouts in a crowdsourcing market from the success rate based on the arrival pattern of workers.