In order to better inform its readers, VRT NWS is experimenting with a new web application, based on artificial intelligence, that allows readers to receive a personalised news offer. The experiment is part of the European research project CPN, led by VRT Innovation, that aims to build a software platform that allows European media organizations to personalise their news content.
To recommend and surprise
This new web application from VRT NWS, called VRT MyNWS, uses a software that learns from your behaviour as a reader and recommends articles based on the content and context that matches your profile. The same technology is also being used to do the exact opposite: recommend articles that could surprise you and go against the so-called “filter bubble”. The experiment aims to measure whether people are better informed when they receive recommendations or not. With the results, the project team can explore future applications, such as push notifications sent at a fitting moment, newsletters that are automatically composed based on your interests, or a personalised video offer.
CPN: a European platform for news personalisation
The test on VRT MyNWS is part of the European CPN project that aims to develop a new recommendation software. For the project, VRT collaborates with technical companies, research institutes as well as media organizations from Belgium, Italy, France, the United Kingdom, Germany, Greece and Cyprus. Since smaller media organizations don’t always have the resources to acquire or develop such software, the VRT completes the request of the European Commission to make the software widely available as a platform. This way, smaller news organizations also have the chance to personalise their offer via the platform and better reach their readers in the current overload of news content. In addition, the project stands out by using data not only to personalise but also to surprise. To make this happen, the software takes into account a diversity of topics, contrasting perspectives and the element of serendipity. The software will be made available as a service, mid 2020.