What is it about?
We trained a recommendation algorithm on twitter data. Then we used large scale estimation of the political attitudes of the user, and we show that the embedding of the algorithm is correlated with the political opinion of users.
Featured Image
Why is it important?
Our work opens the possibility of political explanation methods of recommender systems. It is particularly usefull because he looks at latent embeddings of the algorithm wich are usually overlooked. Understanding the links between the embedding and the political opinions of users can allow us to study polarisation, filter bubbles, radicalisation or other similar phenomenas.
Read the Original
This page is a summary of: Discovering ideological structures in representation learning spaces in recommender systems on social media data, November 2023, ACM (Association for Computing Machinery),
DOI: 10.1145/3625007.3627336.
You can read the full text:
Contributors
The following have contributed to this page
