Harnessing Collective Intelligence on Social Networks

Crowdsourcing is an approach to replace the work traditionally done by a single person with the collective action of a group of people via the Internet. It has established itself in the mainstream of research methodology in recent years using a variety of approaches to engage humans in solving problems that computers, as yet, cannot solve. Several common approaches to crowdsourcing have been successful, including peer production (in which the participants are inherently interested in contributing), microworking (in which participants are paid small amounts of money per task) and games or gamification (in which the participants are entertained as they complete the tasks). An alternative approach to crowdsourcing using social networks is proposed here. Social networks offer access to large user communities through integrated software applications and, as they mature, are utilised in different ways, with decentralised and unevenly-distributed organisation of content. This research investigates whether collective intelligence systems are facilitated better on social networks and how the contributed human effort can be optimised. These questions are investigated using two case studies of problem solving: anaphoric coreference in text documents and classifying images in the marine biology domain. Social networks themselves can be considered inherent, self-organised problem solving systems, an approach defined here as ?groupsourcing?, sharing common features with other crowdsourcing approaches; however, the benefits are tempered with the many challenges this approach presents. In comparison to other methods of crowdsourcing, harnessing collective intelligence on social networks offers a high-accuracy, data-driven and low-cost approach.