Towards a Classification of Content Pollution on Twitter

Twitter and other social media have been heralded as a democratic forum that allows everyone freedom of expression. This view certainly captures an important aspect of Twitter, but more recently, Twitter abounds with content pollution that casts are more sinister view. A remarkable feature of content pollution on Twitter is that it comes in many forms. It involves, among others, spread of SPAM malware, social and other botnets, pump and dump schemes, cybersmear and trolling, fraudulent political campaigning or astroturfing. Recent studies underscore that content pollution on Twitter is a complex and dynamically evolving field. Content pollution typically follows a commercial and/or political agenda, but its overt manifestations are designed to conceal its true purpose. Partly due to standard procedures in scientific research, many studies look only at a narrow section of content pollution. While this approach may lead to detailed results of the focused area, it also implies that fraudulent activities are misunderstood and misclassified. IntheabsenceofaclassificationsystemofcontentpollutiononTwitter,miscommunication can hardly be avoided. For this reason, research and development in this area are not likely to converge. Motivated by the aforementioned discussion, the goal of this paper is to develop an open classification system for content pollution on Twitter. Clearly, such a system needs to be informed by current research into content pollution on Twitter, by insight into existing classification systems and by principles that guide their development. This study seeks to address these lines of work and is structured accordingly. Firstly, forapreliminaryoverview, this study reviews recent work that examines content pollution on Twitter. Secondly, existing classification systems of content pollution on Twitter, e.g., the classification used by the U.S. Securities and Exchange Commission (SEC), are examined. Thirdly, principles that guide classification systems developed in other disciplines are considered. Fourthly, a classification system for content pollution is presented and applied to instances of content pollution on Twitter discussed in the literature. The paper concludes with a discussion on the strengths and shortcomings of the classification system introduced.