Francesca
My rudimentary experiments with linked news items go on. I did some very naive k-means clustering to understand how topics are linked with each other. This cluster here is only composed by articles discussing the FDP "D-Day", with this article being the center of the whole network: https://www.zeit.de/politik/deutschland/2024-11/christian-lindner-ampel-aus-fdp-bundesregierung
FDP: Das liberale Drehbuch für den Regierungssturz

Geheime Sitzungen, Präsentationen, Strategiepapiere: Recherchen der ZEIT zeigen, wie die FDP den Bruch der Ampel wochenlang vorbereitete. Der Name der Operation: "D-Day"

ZEIT ONLINE
Dec 18, 2024 at 3:38pmWeb
Show threadFrancescaDec 18, 2024
Here we have a small cluster about Syria
Show threadFrancescaDec 18, 2024
On its own a small VW cluster. It seems like the topic is not so much connected with the overall political discussion.
Show threadFrancescaDec 18, 2024
This is by no means a very clean or thorough analysis. I still need to clean the data to remove internal links that do not point to articles and I will definetly work out a better clustering strategy. But it's a somewhat interesting entry for my research diary :)
Show threadAleph TwoDec 18, 2024
@framoriniii how do you generate these images?
Show threadFrancescaDec 18, 2024
@alex27 the images are generated and rendered in the browser using JavaScript (D3.js, to be specific)
Show threadAleph TwoDec 18, 2024
@framoriniii go to choice
Show threadZeugsDec 18, 2024
@framoriniii hi, looks cool. I don't get the context what is shown. The Links are links and the nodes are articles,okay. But what are the clusters about. You show the Syria component. Maybe you can add some stubs labels to the nodes or do a topic tf/idf analysis to show the topics of the articles.