An Introduction to Hybrid Human-Machine Information Systems

Abstract

Hybrid Human-Machine Information Systems leverage novel architecturesthat make systematic use of Human Computation by means ofcrowdsourcing. These architectures are capable of scaling over largeamounts of data and simultaneously maintain high-quality data processinglevels by introducing humans into the loop. Such hybrid systemshave been developed to tackle a variety of problems and come withinter-disciplinary challenges. They need to deal with the full spectrumof challenges from the social science standpoint, such as understandingcrowd workers behavior and motivations when performing tasks.These systems also need to overcome highly technical challenges likeconstraint optimization and resource allocation based on limited budgetsand deadlines to be met.In this paper, we introduce the area of Human Computation andpresent an overview of different applications for which Hybrid Human-Machine Information Systems have already been used in the realms ofdata management, information retrieval, natural language processing,semantic web, machine learning, and multimedia to better solve existingproblems. Finally, we discuss current research directions, opportunitiesfor the future development of such systems and their applicationin practice.