Recommendation Systems

Helps Members to exchange anonymized data and retrieve information from third-party recommendation providers.

Traditionally, broadcasters schedule their linear content to suit an assumption of generic group of consumers. However broadcasters online services are becoming more and more relevant, they offer content personalisation based on consumers profile, location, devices and behaviours in addition to the more traditional content genre, time-of-day and channel characteristics. Thus the broadcaster must address a new set of technological challenges. 

Furthermore, vertically integrated Over-the-Top (OTT) providers offer an attractive personalised service. PSM organisations seek to offer the same user experience, while addressing the specific requirements and context of broadcasters.

EBU Technology & Innovation Workplan

Every two years, the EBU develops a roadmap for technology and innovation activities based on the requirements and inputs given by EBU Members. The result of this roadmap is our bi-annual EBU Technology & Innovation Workplan. Strategic programmes and project groups are set up to focus on specific areas of interest. To access the latest Workplan, click here.

EBU Project Group on Recommendation Systems

The topic of Recommendation Systems is part of the EBU's Strategic Programme on Broadcaster Internet Services.

Main activities:

Recommendations involve complex distributed systems and require a high level of tuning in order to best engage the audience. There are three basic input components of broadcast content recommendations:
  • Editorial, which depends on the culture, language and editorial choices of the media.
  • Automated, which is mainly based on algorithms such as collaborative filtering, data mining, clustering. This will for instance provide archive crawling and   discovery as well as clustering of content and audience.
  • Social, which is based on recommendations from friends. However, this requires the existence of the concept of community in the broadcaster’s user experience.
  • EBU Members can leverage resources, thus reducing their costs, by running a common system to provide automated and/or editorialised recommendations. 
Also, integration of third-party systems is often expensive. The RecSys Group aims provides a standard way of exchanging anonymized data and retrieving information from third-party recommendation provider.
Finally, any recommendations systems needs to take into account the activity of a user in real-time in order to provide the most relevant content. In order to achieve the necessary requirements for high-availability, scalability, security as well as the need for high volumetric storage, state of the art elements run by the open source community - like Apache, Hadoop and Spark - are leveraged.

​Join us

If you are interested in Recommendation Systems, join our group on this topic and participate in the discussions. Access for EBU members only.



Related topics