Do you study political networks and use AI-tools to gather the necessary data? Join our workshop and discuss your approach and findings with like-minded peers!
 

Networks or connections among political actors play a crucial role in politics: alliances and enmities shape national and international conflicts, patronage ties enable corruption and career advancement, and the spread of rumors along informal ties can topple governments. But gathering information on such networks is labor-intensive.
 

Fortunately, recent advances in natural language processing, especially the development of large language models, have introduced ways to automate and greatly accelerate the inference of networks from text. What remains uncertain, however, are LLMs’ reliability and replicability - and what completely new social network questions can be answered using these tools.
 

For our workshop, we are particularly interested in:

​

1. Studies that explore new political network questions enabled by AI-tools or that revisit classical questions using such tools.

2. Analyses that either compare different data-gathering approaches to the same case or apply the same approach to different cases to test its generalizability.

3. Innovative ways to combine human and AI-inputs in the data-gathering process

4. Novel ways to ensure a) replicability, in particular for proprietary AI-tools, and b) test the reliability or external validity of networks constructed, i.e. construct ground-truth data sets

5. Methods for improving LLM effectiveness and scalability in extracting political networks

 

We welcome research on both open-source and proprietary LLMs and other AI tools. While the focus will be on empirical and methodological studies, we are also open to conceptual and theoretical work, in particular on the limitations of this approach for different contexts, types of actors and networks. You may also submit in-depth research into parts of the data-gathering process, such as identifying the relevant entities, establishing the type or sentiment of the relationship, or gathering CV or other data used to construct networks.

​

We plan to make space for shorter presentations to get feedback on specific research ideas (for a PhD-thesis, a grant application, or a paper) - please indicate such proposals by beginning your submission with “Research proposal:”

​

Finally, also get in touch with us directly if you have developed tools, packages or programs relevant to this endeavor and would like to give a hands-on demonstration to this audience.

​

Note that this is an in-person event, but that ECPR offers travel grants to students and early career scholars from Full ECPR member institutions.

​

If you want to join the workshop, submit an abstract of no more than 500 words by December 10, 2025, here: https://ecpr.eu/Events/Event/WorkshopDetails/16809. For questions, please contact Franziska Keller (franziska.keller@unibe.ch) or Yuequan Guo (yuequan.guo@wzb.eu).