Georg Ahnert
LightNews LightNews
banner
wanlo.bsky.social
Georg Ahnert
@wanlo.bsky.social
200 followers 410 following 15 posts
PhD Student in Social Data Science at University of Mannheim | LLMs and Surveys | georgahnert.de
Posts Replies Media Videos
wanlo.bsky.social
Here‘s some of the slides 👇 bsky.app/profile/mstr...
July 24, 2025 at 4:20 PM
wanlo.bsky.social
The 15.07 train has a 30 min delay now but the landscape‘s quite pretty ;)
The Swedish countryside as seen from a moving train, with a lake, a red and white house, and some cows.
July 20, 2025 at 4:19 PM
wanlo.bsky.social
Thanks :) We have a BERT-based baseline model that labels individual tweets—but I agree, would be a very interesting comparison now that LLMs can increasingly handle super long contexts!
June 27, 2025 at 9:53 AM
wanlo.bsky.social
Thanks a lot for the shoutout! Would be happy to talk about this and other ongoing projects on social simulation at #ICWSM next week 🙂
wanlo.bsky.social Georg Ahnert @wanlo.bsky.social · Jun 16
Excited to present our paper with @maxpe.bsky.social, @dgarcia.eu, and @mstrohm.bsky.social next week at #ICWSM! ✨

We extend social simulation with LLMs to a longitudinal setting by fine-tuning Temporal Adapters—here's how: 🧵
A lineplot that shows how scared people in the UK were over time, during the first COVID-19 lockdown. Our method (Llama 3 Temporal Adapters) produces similar estimates of as the survey data gathered by YouGov.
June 16, 2025 at 8:29 AM
wanlo.bsky.social
Our conclusion: Temporal Adapters enable longitudinal analyses of affect aggregates from social media data by temporally aligning LLMs. ⏱️

Read the full paper: ojs.aaai.org/index.php/IC...
Extracting Affect Aggregates from Longitudinal Social Media Data with Temporal Adapters for Large Language Models | Proceedings of the International AAAI Conference on Web and Social Media ...
ojs.aaai.org
June 16, 2025 at 8:23 AM
wanlo.bsky.social
Results: From several collective emotions and public opinion, our longitudinal estimates show a strong positive and significant cross-correlation with survey data gathered by YouGov directly from human participants.
Our estimates with Llama 3 Temporal Adapters show a strong positive and significant correlation with collective frustration, fear, boredom, and sadness. Our results vary strongly between emotions, but they are in line with a baseline method's estimates. We also apply our method to the extraction of public attitudes towards Boris Johnson as a prime minister and towards the National Healthy Service, were we similarly find positive cross-correlation with survey data for some but not all answer options.
June 16, 2025 at 8:23 AM
wanlo.bsky.social
Method: We gather weekly text data from a panel of Twitter users and fine-tune Temporal Adapters for Llama 3 8B with it. 🦙 We then prompt Llama with established survey questions, one week at a time, to extract longitudinal affect aggregates.
Overview of our method that shows how each week's Twitter data is used to train a separate Temporal Adapter, and how a weekly affect aggregate is then obtained from the LLM's token probabilities.
June 16, 2025 at 8:23 AM