Published September 10, 2025
| Version v1
Presentation
Open
Building a human-centered Large Language Model dataset from longitudinal QESB (2010 to 2024) data
Authors/Creators
- 1. GESIS Leibnitz Institute for the Social Sciences
-
2.
GESIS - Leibniz-Institute for the Social Sciences
- 3. University of the West of England, Bristol
-
4.
University of Dundee
-
5.
University of Würzburg
Description
The presentation showcases the successful pilot project, the QESB LLM, a pioneering initiative focused on building a human-centered Large Language Model (LLM) dataset using longitudinal Qualitative Election Study of Britain (2024) data. While current political science applications of LLMs focus on social media or news content, the QESB LLM will offers a richer alternative: longitudinally consistent data that captures how ordinary citizens talk and think about British politics in their own words. By combining qualitative insight with machine learning tools, the project aims to support both social science discovery and ethical LLM development.
Files
Building a human-centered Large Language Model dataset from-1.pdf
Files
(335.2 kB)
| Name | Size | Download all |
|---|---|---|
|
md5:1fd43af104f1409dd2275818bd909a90
|
335.2 kB | Preview Download |
Additional details
Dates
- Available
-
2025-09-10