22 min listen
Machine Learning the Facebook URLs Dataset to Study News Credibility, with Dr. Tom Paskhalis
Machine Learning the Facebook URLs Dataset to Study News Credibility, with Dr. Tom Paskhalis
ratings:
Length:
43 minutes
Released:
Aug 21, 2022
Format:
Podcast episode
Description
Dr. Tom Paskhalis, Assistant Professor in Political and Data Science at Trinity College Dublin, shares his research on applying machine learning to the Facebook URLs Dataset from Social Science One. The project develops a model to label whether a news domain is credible or not based on Facebook interactions data. We discuss the Facebook URLs dataset, what types of machine learning techniques were applied to it, and how the model performed across the US and EU countries.
Released:
Aug 21, 2022
Format:
Podcast episode
Titles in the series (100)
Social Media and Political Youth Organizations in Denmark, with Emilie Demant: Emilie Demant, social media coordinator for Venstres Ungdom, shares her insights into how a Danish political youth organization is using social media to engage young voters with politics. We discuss how Facebook, Snapchat, Instagram, and Twitter are each used differently to communicate politics with young Danes, as well as what types of user-generated content Emilie receives when managing these social media accounts. Emilie highlights the visual element of social media by stressing that memes, GIFs, and videos drive the most engagement on social media, and here digital marketing and graphic design play a key role. We also discuss the differences between a youth political organization and the parent political party, Venstre, and what that means for their social media use. Although exhibiting different rules of political communication on social media (especially on Snapchat), interestingly, both Venstre and Ve by Social Media and Politics