Toward Detecting Conspiracy Language in Misinformation Documents

Abstract

Recent events have illustrated the danger of online conspiracy theories to cause harm in the real world. In this research in progress, we focus on machine learning techniques to differentiate conspiracy information from other forms of misinformation. Our results demonstrate that conspiracy language is differentiable and suggest there may exist features of conspiracy language independent of the conspiracy topic. We also present a direction for future work to better understand the unique features of conspiracy language and how they may be used to enhance machine learning techniques.