Quality Assessment of Peer-Produced Content in Knowledge Repositories Using Big Data and Social Networks

Abstract

This research provides a method for quality assessment of peer-produced content in knowledge repositories using a complementary view of collaboration. Using the definition of collaboration as the action of working with someone to produce something, we identify the aspects of collaboration that the present research on online communities does not consider. To this end, we introduce and define the concept of implicit collaboration and then identify two dimensions and four possible areas of collaboration. In each area, we identify the relevant social network that captures collaboration. Using customized measures on each of the networks that capture various aspects of collaboration, we quantify the utility of implicit collaboration in assessing article quality. Experiments conducted on the complete population of graded English language Wikipedia articles show that all the identified measures improve the predictive accuracy of the existing models by 11.89 percent while improving the class-wise precision by 9-18 percent and the class-wise recall by 5-26 percent. We also find that our method complements the existing quality assessment approaches well. Our research has implications for developing automated quality assessment methods for peer-produced content using big data and social networks.