Recommendations Explained: Towards Transparency and Fairness for Various Stakeholders

Recommender Systems operate in more complex scenarios than expressed in the simplified objective of “maximizing the consumer’s experience”. Platforms typically serve multiple stakeholders with different, possibly competing and conflicting interests. This doctoral project investigates the needs, understandings, and expectations they have of Recommender Systems, e.g. with respect to transparency and fair outcomes and opportunities. The candidate develops Recommender Systems that cater to these different needs and develop Explainable AI approaches able to communicate relevant aspects to different user groups, negotiating between transparency, privacy and business interests. The primary use case will be established within the area of music streaming, possibly with outreach to other domains.

Peter Knees – Faculty of Informatics, TU Wien

Verena Dorner – Dept. of Information Systems and Operations Management, WU Wien

Mark Coeckelbergh – Department of Philosophy, University of Vienna