Public defence in Computer Science, MA Henna Paakki

Title of the thesis: Asymmetric Conversational Strategies: Methods for Detecting Manipulative Online Trolling

Doctoral student: Henna Paakki
Opponent: Professor Asta Zelenkauskaite, College of Arts and Sciences, Drexel University, USA
Custos: Professor of Practice Nitin Sawhney, Aalto University School of Science, Department of Computer Science

Trolling is increasingly used to influence and disrupt conversations on social media. Particularly harmful is trolling that aims to influence political and societal discussions, or to systematically target and harass discussion groups. This research investigated the features that different types of trolling may have in common on different discussion platforms. The goal was to develop more effective tools for automated trolling detection. Previous research on trolling detection has often focused on the characteristics of individual messages, for example their sentiment. This study, on the other hand, emphasizes the interactive nature of trolling.

This dissertation offers new information about conversational features that are important for trolling detection. Trolls often violate the unspoken rules of conversation, such as expectations related to responding, to manipulate others. The main contribution of this research is to develop an effective trolling detection model based on analyzing conversational features – one that surpasses earlier trolling detection models in performance. This research shows that even indirect and deceptive forms of harassment such as trolling can be identified using machine learning. The developed model can be used to identify trolling on social media platforms, to support moderation, and to prevent harmful influence and harassment in online spaces.

Key words: trolling, social media, automated detection, online manipulation, harassment, conversational features

Thesis available for public display 10 days prior to the defence at: https://aaltodoc.aalto.fi/doc_public/eonly/riiputus/ 

Contact information:

Doctoral theses of the School of Science: https://aaltodoc.aalto.fi/handle/123456789/52