Dropout Prediction in Crowdsourcing Markets

TL;DR

This paper explores predicting worker dropout in crowdsourcing markets by analyzing success rates and arrival patterns, aiming to improve task completion stability.

Contribution

It introduces a method to predict worker dropout based on success rates and arrival patterns, addressing a key challenge in crowdsourcing stability.

Findings

Dropout correlates with success rate decline.

Arrival patterns can predict dropout likelihood.

Predictive model can enhance crowdsourcing management.

Abstract

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.