Media Observatory Initiative

Consumers are confronted with a deluge of digital information each day, and it is a difficult task to select the most appropriate, credible, and authentic news sources. The situation is further complicated by the fact that many of those sources lead to false news. That leads to a trust deficit between news sources and consumers. The Media Observatory Initiative seeks to address that challenge by creating a web platform to capture, process, and visualize the Swiss and global news landscape. Its transparent quality will help consumers assess the reliability of news sources and information. On the other hand, the platform will allow journalists to keep track of the reaction of audiences online and effectively identify and address controversial or polarizing issues.

There are three main cornerstones in our approach:

  • We propose a model of the media landscape that automatically maps news sources based on their selection of subjects.
  • By tracking sources’ evolution over time, we identify driving forces, from the influence of ownership to large-scale content diffusion patterns.
  • We produce a dynamic map of the media landscape over the last 3 years, based on coverage from 500 million articles shared by around 8000 sources.

Based on these findings, we will release an open and interactive map that can be used by viewers and journalists alike to dig into the patterns that influence the media ecosystem. Our approach compares the coverage of events across news entities, which produces a similarity map where distances between channels represent the similarity of their coverage. Technically speaking, our method borrows from the field of personalization with a set of techniques commonly used in recommender systems, such as those used on Amazon or Netflix.

Our approach has potential for news monitoring applications and investigative journalism by shedding light on important changes in programming induced by mergers and acquisitions, policy changes, or network-wide content diffusion. A key observation in our analysis is the convergence of a group of sources towards a specific position. We propose an automated method to detect this phenomenon by introducing the notion of attractors. The algorithm identifies regions of the map toward which many sources converge. Once identified, the attracted news sources, which are converging towards an attractor, can be monitored.

The research team is working in partnership with Swiss daily newspaper Le Temps. The project is funded by the Initiative for Media Innovation (IMI).