Identifying and Evaluating Methods for Testing the Impact of Recommender Systems

Ofcom has a requirement for a research project to investigate the use and impact of recommender systems by online services. We are seeking to develop our understanding of the key functionalities and features of in-scope services. One of these is the use of algorithmically-driven recommender systems, which are used to serve up content both on search services (e.g. search engine sites) and user-to-user services (e.g. social media platforms and video sharing platforms). We are seeking to collect evidence on how these systems function at a technical level, where and how they are used in practice, the risks they present to users, and how these risks can be mitigated. This research will inform the development of the regulatory documents that we are required to prepare under the Online Safety Bill to assist in-scope services in understanding how they can comply with their duties, including the Codes of Practice, the Register of Risks and Risk Profiles that will come out of Ofcom's risk assessments, and Industry Risk Assessment Guidance.