Actantial Networks

Connections can tell a story that individual observations cannot. Listeners will learn about Actantial Networks—graph-based representations of narratives where nodes are actors (such as people, institutions, or abstract entities) and edges represent the actions or relationships between them. The one who will present these networks is our guest Armin Pournaki, a joint PhD candidate at the Max Planck Institute and Sciences, who specializes in computational social science, where he develops methods to extract and analyze political narratives using natural language processing and network science.

Guest

Armin Pournaki: I am a PhD candidate in the groups of Jürgen Jost (MPI MiS) and Thierry Poibeau (Lattice). Together with them, Eckehard Olbrich (MPI MiS) and Jean-Philippe Cointet (médialab), I explore computational approaches to language and discourse analysis. More precisely, I am interested in computationally extracting (political) narratives from raw text and empirically investigating their role in phenomena like polarization and issue alignment. This work is embedded in the EU Horizon project SoMe4Dem, in which we aim to provide stronger empirical evidence for the impact of social media on political debates and democratic processes. Previously, I was part of the ODYCCEUS project, where I wrote my master's thesis on opinion-dynamics based approaches to community detection and contributed by designing opinion observatories like the twitter explorer. Before that, I wrote my bachelor's thesis in theoretical physics at the Technische Universität Berlin on synchronization patterns in modular neuronal networks under the supervision of Philipp Hövel.