Seminar 1: Digital Direct Democracy – Enhancing Citizen Participation through Human-AI Collaboration

Speaker: Daan van der Weijden, University of Zürich, Switzerland

Date and time: 10 – 11am Wednesday 7 May 2025
Room: 47A-141 – Sir James Foots Building
Host: Prof Gianluca Demartini

Abstract: In an era were democracy as we know it is at risk, direct democracy systems oaer a potential pathway to revitalize citizen participation. This research explores the integration of large language models LLMs into Switzerland’s direct democratic process, examining how these technologies can help and guide citizens across three key phases: deliberation, co-creation, and voting. Our work investigates how LLMs can lower barriers to democratic engagement by making complex policy information more accessible, facilitating diverse citizen input in legislative drafting, and guiding them in the deliberative process.

Bio: I am a computer science PhD student with a passion for using technology to improve democratic decision-making processes. My research focuses on the use of artificial intelligence and natural language processing in digital deliberative democracy, where citizens come together to discuss and make decisions through online platforms. With my expertise in these fields, I am dedicated to creating innovative solutions that can help people engage in more meaningful and eaective democratic discourse.

Seminar 2 – Simulating Social Media Using LLMs (Under construction)

Speaker: Azza Bouleimen, University of Zürich, Switzerland

Date and time: 11am – 12pm Wednesday 7 May 2025
Room: 47A-141 – Sir James Foots Building
Host: Prof Gianluca Demartini

Abstract: To mitigate harms on social media (misinformation, hate speech, polarization), online platforms implemented a set of intervention policies. However, research highlights the limited impact of these policies, worse, some had the opposite eaect. Why are these policies failing at curbing undesired behaviour on social media? Is there a better approach to design such interventions? In this context, simulating social media using LLMs appears to be a promising approach for designing, testing and tuning diaerent policies before deploying them. However, to which extent are LLMs capable of mimicking the collective behaviour of humans on social media? In my talk, I will attempt to answer some of these questions and ask many others.

Bio: I am a PhD candidate at the University of Zurich. I joined the Social Computing Group under the supervision of Prof. Anikó Hannák in 2022. My research focuses on alternative approaches to designing social media intervention policies. I am currently studying the collective behaviour of LLMs and assess their capabilities at mimicking human behaviour on social media. However, I remain enthusiastic about other research interests. For instance, in the future, I will be very curious to investigate the « sociology » of LLM groups, and study whether LLMs demonstrate similar social behavioural patterns to humans or not.

Seminar 3 – Towards Ontology and Knowledge Graph Lifecycles

Speaker: Romana Pernisch, Vrije Universiteit Amsterdam

Date and time: 1pm Wednesday 7 May 2025
Room: 67-343 – Priestley Building
Host: Prof Gianluca Demartini

Abstract: The lifecycle of ontologies and knowledge graphs plays a crucial role in managing structured knowledge in evolving domains. This talk presents a systematic approach to their engineering and evolution. First, I will discuss the methodological framework for designing and maintaining ontologies and knowledge graphs. I will then talk about ontology evolution, and present some analyses of impact on downstream applications which segways into the impact on the KG engineering process when ontologies evolve. I will touch on Knowledge Graph construction dependencies and also present the first progress in automatically updating RML mappings based on ontology changes. By addressing these interconnected challenges, this talk provides a comprehensive perspective on sustaining and advancing knowledge-driven systems.

Bio: Romana Pernisch is a postdoctoral researcher at the Vrije Universiteit Amsterdam, where she is part of the Knowledge in Artificial Intelligence group (https://kai.cs.vu.nl/) and the Discovery Lab (https://discoverylab.ai/). Her research primarily focuses on Knowledge Graph Evolution, aiming to enable Knowledge Graph lifecycles through innovative Software Engineering approaches. Romana has made significant contributions to the field, earning a ISWC Best Poster Award in 2019. She has also been actively involved in the academic community, serving as the Poster and Demo Chair for the 23rd International Semantic Web Conference (ISWC 2024) and the 18th International Conference on Distributed and Event-based Systems (DEBS 2024).