RecTurk: Constraint-based Recommendation based on Human Computation

The development of constraint-based recommender applications is still challenged by the knowledge acquisition bottleneck. The management of the underlying constraint sets is often accompanied by additional eorts related to the information exchange between knowledge engineers and domain experts. In this paper we introduce the RecTurk research prototype which supports the development of constraintbased recommender applications on the basis of Human Computation concepts. Thus we substitute complex knowledge engineering tasks with simple micro-tasks that can be performed by persons without experiences in constraintbased recommender application development. We introduce the basic concepts currently integrated in RecTurk and report the results of a rst user study that evaluated the applicability of RecTurk in three item domains.