Crowd, a platform for the crowdsourcing of complex tasks

BDA 2014, Gestion de données - principes, technologies et applications, Autrans, France

Conference was on October 2014
Accepted as regular paper

Authors

Ahmad Chettih, David Gross-Amblard, David Guyon, Erwann Legeay and Zoltan Miklos

Abstract

Crowdsourcing is an emerging technique that enables to involve humans into information gathering or computational tasks. With the help of crowdsourcing platforms a group of participants (workers) can solve otherwise difficult problems. While the existing, generic crowdsourcing platforms such as Amazon Mechanical Turk have been used to address various challenges, they only support simple questions that we consider as basic tasks. In this demo we present Crowd, a platform where one can submit a workflow, which is a composition of such basic tasks. Crowd also supports a simple skill-management mechanism":" each basic task is annotated with expertise tags. Upon the validation of completed tasks, the worker's expertise is updated according to these tags. The worker's expertise can later be used for better task selection. Crowd is implemented in Python/Django, and can be used both on the Web or on mobile devices.