Measuring the Quality of Edits to Wikipedia

Abstract

Wikipedia is unique among reference works both in its scale and in the openness of its editing interface. The question of how it can achieve and maintain high-quality encyclopedic articles is an area of active research. In order to address this question, researchers need to build consensus around a sensible metric to assess the quality of contributions to articles. This measure must not only reflect an intuitive concept of "quality," but must also be scalable and run efficiently. Building on prior work in this area, this paper uses human raters through Amazon Mechanical Turk to validate an efficient, automated quality metric.