Call for Participation: Internet Research with Foundation Models?
CAT4SMR is convening its closing workshop on the exploration of generative methods in internet research on September 3, 2024 in Amsterdam. Situated within the Digital Methods Initiative, CAT4SMR has been working on tools for capturing and analyzing social media data, such as 4CAT and the YouTube Data Tools, but also on how to develop and implement new and innovative research methods.
As the incorporation of large pre-trained models (sometimes referred to as “foundation models”, including Large Language Models, Large Vision Models, etc.) into social sciences and humanities (SSH) research moves from novelty to standard, the need to exchange, evaluate, and critique is becoming pressing. This one-day workshop will serve as a forum for investigating how these advanced computational techniques are being integrated into digital research, what opportunities, limitations, and pitfalls they imply, and what this means for academic research.
Our conversations will center around several themes:
- The evolution of digital methods in light of foundation models, exploring how tools like LLMs and LVMs augment and transform established methodologies and introduce new ones into digital research, from thematic textual and visual analysis to the study of misinformation;
- The critical examination of best practices in the deployment of LLMs or LVMs, from the art of effective prompting to the nuances of model selection, fine-tuning, and evaluation, all within the context of internet research;
We are particularly interested in fostering discussions about new epistemologies, ethical implications, and practical issues related to the use of these models in research. This includes the careful examination of models’ black-boxed nature and their (potential) imprint on research, as well as the broader ethical landscape that researchers navigate.
We welcome short proposals (about 500 words) for brief presentations on empirical research projects, methodologies, tools, or critiques aligned with our workshop themes, especially those adopting experimental, explorative, or speculative approaches. Potential topics include, but are not limited to:
- interesting new methods, advancing visual and textual analysis;
- effective strategies for employing pre-trained models in internet research;
- taming black boxes: observing, testing, and deploying pre-trained models for digital research;
- the evolution of interpretive practices with the integration of LLMs;
- the role of pre-trained models as surrogate expert readers;
- the delegation of interpretative agency to pre-trained models;
- reflections on the role of pre-trained models in, or next to, qualitative and quantitative research;
- challenges and opportunities in the ‘traceability’ of digital actors and research procedures;
- researching platformed models;
- platform research with pre-trained models and their (mis)uses for medium-sensitive studies;
This will be a one-day meeting with presentations and ample room for discussion. We hope to make this an inspiring event by bringing together scholars and practitioners from diverse disciplines, resulting in fruitful collaborations. We have some ability to fund transport and accommodation, and will provide lunch and dinner.
Deadline for submissions: June 7, 2024
Submit proposals to: cat4smr [diddalidoo] list.uva.nl
Notice of acceptance: June 21, 2024
Workshop date: September 3, 2024
Location: Amsterdam City Centre
Organizers: Erik Borra, Sal Hagen, Stijn Peeters, Bernhard Rieder