Dreaming the Machine: Monitoring and Datafication of Hate Speech With Machine Learning

• In Event: Methodological Challenges of Hateful and Problematic Content Detection Online

Abstract

During the past few years societies have recognized a growing problem of discriminating, hateful speech online, often targeting specific groups and minorities in special. In this paper we reflect the experiences gathered during a research-NGO collaboration conducted during municipal elections in a European country in 2017, with the aim of automatically monitoring candidates’ social media updates for hate speech and using machine learning to filter potential hate speech. The collaboration offered a unique view to investigate, first, how hate speech emerges as a technical problem. An adequately well-working machine learning solution was developed in the project, but the automated approach requires heavy simplification, such as using rudimentary scales for classifying hate speech, which in reality has several different tones and varieties. Second, the setting allowed us to study the affective responses such as hopes, dreams, and fears related to the machine learning technology by the project participants.

Authors

• Salla-Maaria Laaksonen, U of Helsinki

• Reeta Pöyhtäri, University of Tampere

• Matti Nelimarkka, Helsinki Institute for Information Technology

• Jesse Haapoja, U of Helsinki

• Teemu Kinnunen, Futurice Oy