Data Skeptic Actantial Networks
27 snips
Jun 1, 2025 In this discussion, Armin Pournaki, a Joint PhD candidate at the Max Planck Institute, unfolds the concept of Actantial Networks. He reveals how these graph-based structures can dissect political narratives, showcasing how conflicting stories arise around events like COVID-19 and the war in Ukraine. Pournaki also highlights how natural language processing helps visualize social media discourse, aiding in understanding polarization and narrative persuasion. His insights transform the way we perceive political communication in a divisive landscape.
AI Snips
Chapters
Transcript
Episode notes
Narratives Represent Political Reality
- Narratives represent political reality through actors, goals, and events causing world state changes.
- These can be reconstructed from text to identify causal connections and narrative signals in discourse.
AMR Enables Sentence Graphs
- Abstract Meaning Representation (AMR) translates sentences into directed acyclic graphs for actor-action-role extraction.
- This graph structure enables easy querying of actors and relationships to build higher narrative representations.
Actantial Networks Capture Narrative Conflict
- Actantial graphs use actors as nodes and encode relationships as supportive or conflictive edges.
- Edge scores from verb categorization summarize narrative relationships, allowing detection of conflicting narratives.