Manuel Tonneau

About

Manuel is a PhD candidate in Social Data Science at the Oxford Internet Institute of the University of Oxford. During his time at the Weizenbaum Institute, Manuel will further develop his PhD research on cross-geographic disparities in the performance of automatic hate speech detection and the enforcement of hate speech moderation on Twitter. To do this, he collected a large dataset of all tweets posted during one day and had large random samples annotated as hateful or not in eight languages (Arabic, Bahasa, English, French, German, Portuguese, Spanish and Turkish). He will now use this annotated dataset to evaluate state-of-the-art automatic hate speech detection methods in real-world conditions and inspect the proportion of all hate that was moderated by the platform. With this project, Manuel hopes to enhance platform accountability and improve moderation practices globally.

Manuel’s PhD research is focused on the moderation of online harmful content. Specifically, he studies the extent to which such moderation may treat users unequally across geographies, from unequal automatic detection performance to unequal moderation enforcement through a potential misalignment between user preferences towards moderation and platform moderation rules. Prior to his PhD, Manuel worked for the World Bank and the Centre Marc Bloch extracting policy-relevant information from large social media datasets in Minority World settings using natural language processing and network science. He holds an Engineering degree in Statistics and Economics (eq. to MSc) from ENSAE Paris as well as an MSc in Economics from Humboldt-Universität zu Berlin.

Manuel’s work has appeared in leading computer science proceedings (ACL, ICWSM, GoodIT). His doctoral work is generously funded by a Global Merit Award from the Shirley Scholars Fund. Beyond his PhD, Manuel consults for the Development Impact Unit of the World Bank and is affiliated with the Open Networks and Big Data Lab at New York University.