This project aims to visualize the semantic similarity of news articles from multiple perspectives by going beyond standard thematic similarity, and explore whether polarity can be visualized in articles covering a given media event. Such a tool can raise awareness on the issue of polarity among media content producers and aggregators. In addition, by exploring the “media space” of a given event, users can better understand if they are being caught in filter bubbles, while not giving up on the potential for content personalization. Finally, it also helps journalists to position their articles and navigate among others in the “media space”.

CommPass is an extension to the “Media Observatory Initiative” project that led to the creation of the NewsTeller platform. NewsTeller captures, processes and references over 1.5 million news articles per month in multiple languages with improved context. This project seeks to provide a more intuitive way of exploring the media landscape and thus enhance the NewsTeller platform with new visualization tools.

CommPass Embedding Projector demo

Watch on YouTube

Keywords :
observatory, media space, news ecosystem, visualization, polarization

3 months

People involved

Principal investigator

Prof. Karl Aberer (EPFL)



Rémi Lebret (EPFL)


Software engineer

Stéphane Massonnet (EPFL)

Academic institution

Distributed Information Systems Laboratory – LSIR (EPFL)


This project started in October 2022 and has been completed