ODI Fridays: How the BBC builds public service recommenders with data

The BBC is one of the world’s leading broadcasters, producing a large amount of content including video, audio and text, spanning topics such as news, sport, education, and entertainment. In order to fulfil its public service remit, the BBC must serve all its audiences, providing each audience member with the most relevant and engaging content for them. Up to now, manual curation has been the main approach used by the organisation to achieve that. However, that comes with some challenges: it is unable to scale and to be tailored to each user. Datalab has followed an approach based on cross-team collaboration and on the concept of public service AI to address these challenges. In his talk, Alessandro reflects upon this approach and the lessons learnt in the years since the creation of Datalab. About the speaker Alessandro Piscopo is a principal data scientist at the BBC, in a team called Datalab. He works on the development and deployment of public service recommendations algorithms across the organisation, to help BBC audiences find the most relevant and engaging content. During his PhD, obtained in 2019 at the University of Southampton, Alessandro studied the effects of socio-technical dynamics on data quality in collaborative knowledge engineering projects. His current interests include online collaboration, knowledge graphs, and fairness and ethics in AI.