A joint University of Southampton and Microsoft Research Cambridge team, lead by the ORCHID researcher Matteo Venanzi, were one of the winners of the Crowdsourcing at Scale 2013 Shared Task Challenge. The challenge involved computing accurate aggregations of tween sentiment judgments for large crowdsourced datasets collected by Google and Crowdflower. The Soton/MSR team designed a Bayesian multi-class classification crowdsourcing model for information-rich settings, which combines text features with numeric labels, and can be scaled to large datasets. The results on the competition are featured on the Crowdflower blog.
https://crowdflower.com/blog/2013/12/crowdsourcing-at-scale-shared-task-challenge-winners/