Data-driven recommendations in a public service organisation

Abstract

The BBC is one of the world's leading broadcasters, producing a large amount of content for different audiences. Data-driven recommendations are a successful approach to increase user engagement providing tailored content and personalising their experience. However, concerns have been raised with regards to their effects on diversity and reinforcement of existing bias. Addressing these concerns is especially important for the BBC, whose values include trust, diversity, and impartiality. This position paper lays out the strategy followed by the BBC to develop automated recommendation systems, presenting our approach to create accurate, fair, and responsible recommendation systems.