The Project

Detecting macropatterns in global media content

Every day hundreds of millions of people read millions of news items, in dozens of languages. These news, coming from newswires, newspapers, broadcast or social media, play a crucial role in shaping public opinion and informing the democratic process, as well as influencing markets and industry.

Each item forming this enormous stream of information has been
written, or chosen, by an editor, with the purpose of satisfying the information need of the intended audience (as well as a variety of other constraints).

The contents of the news mediasphere form a dynamic and complex
landscape, but not one without patterns and regularities . Social scientists have been interested in understanding them, as they both reflect and influence patterns in public opinion and society.

This study has the purpose of automating the analysis of news content, by deploying state of the art methodologies from Artificial Intelligence, Data Mining and Web Technologies. We analyse millions of articles from hundreds of outlets and tens of languages, to understand what laws influence the contents of the media system. We have already detected several such strong patterns.

As many aspects of scientific investigation have now been automated, e.g. molecular biology, some parts of the social sciences may be next. This study is aimed at exploring how much we can learn of media content patterns by large scale automated content analysis.