Perceived Task Similarities for Task Recommendation in Crowdsourcing Systems

Abstract

Crowdsourcing platforms support the assignment of jobs while relying on the workers' search capabilities. Recommenders can support the workers' decisions to improve quality and outcome for both worker and requester. A precedent study showed, that many workers expect to get tasks recommended, which are similar to previously finished ones. In order to create genuine task recommendation, similarities between tasks have to be identified and analyzed. Therefore, this work provides an empirical study about how workers perceive task similarities. The perceived task similarities may vary between workers with different cultural background and may depend e.g. on the complexity, required action or the requester of the task.