Our aim in this study is to support the research community in exploring the dynamics of how misleading narratives are developed and spread on social media during the European elections (UK general election as an example).
We present example tweets and corresponding super-narratives and narratives from our dataset.
Our contributions are as follows:
1. We develop a multi-level codebook covering narratives surrounding the European Elections.
2. We introduce UKElectionNarratives, the first Twitter dataset for detecting misleading narratives during the UK general Elections in 2019 and 2024.
3. We present benchmarking results on UKElectionNarratives, and release our source code for reproducibility and facilitating research on the task.
4. We explore how Large Language models perform under zero-show and in-context learning setups.
5. We explore how our models perform under in-domain and cross-domain setups.
The rest of this paper is organized as follows: we review the literature and we present our dataset construction approach.
HAOUARI Fatima;
SCARTON Carolina;
FAGGIANI Nicolò;
NIKOLAIDIS Nikolaos;
KOTSEVA Bonka;
ABU FARHA Ibrahim;
LINGE Jens;
BONTCHEVA Kalina;
2025-12-10
AAAI Press
JRC142073
1-57735-900-3 (online),
3067-1515 (online),
https://ojs.aaai.org/index.php/ICWSM/article/view/35950/38104,
https://publications.jrc.ec.europa.eu/repository/handle/JRC142073,
10.48550/arXiv.2505.05459 (online),
| Name | Country | City | Type |
|---|
This document is only visible at the Commission level.
You are not authorized to publish or distribute it outside the European Commission.
This is a public document. You can share this publication.
Datasets
| ID | Title | Public URL |
|---|
Dataset collections
| ID | Acronym | Title | Public URL |
|---|
Scripts / source codes
| Description | Public URL |
|---|
Additional supporting files
| File name | Description | File type |
|---|