Title: The next billion: design choices and stakeholder outcomes in multi-stakeholder RecSys

Opening keynote by Rishabh Mehrotra, ShareChat.

Abstract

Recommender systems shape the bulk of consumption on digital platforms, and are increasingly expected to not only support consumer needs but also benefit creators and suppliers by helping them grow their audience, while supporting platform economics to ensure long-term health and sustainability of the platform. In this talk, we view the recommendation stack from a multi-stakeholder lens and scrutinize how the intricate design choices across the RecSys stack have an impact on the outcomes across different stakeholders. We start by dissecting the heterogeneities present in such ecosystems, spanning from user specific to content and creator specific to platform intricacies. Diving deep into the modern large-scale RecSys stack, we illustrate how each component makes specific encoding decisions, and these choices being far from innocuous, manifest second-order effects that ripple across the platform’s ecosystem, differentially affecting a wide range of stakeholders. Finally, we will highlight the importance of crafting effective interventions while being cognizant of stakeholder incentives: can platforms evolve responsively without jeopardizing their bottom line? The talk presents a view from the trenches aimed at nudging the community towards understanding and addressing FAccT challenges in multi-stakeholder platforms under realistic industrial considerations.

About Rishabh

Rishabh Mehrotra currently works as a Director of Machine Learning at ShareChat. His current research focuses on efficient recommender design for feed ranking, multi-objective modeling of recommenders and creator ecosystem. Prior to ShareChat, he was an Area Tech Lead and Staff Scientist/Engineer at Spotify where he led multiple ML projects from basic research to production across 400+ million users. Rishabh has a PhD in Machine Learning from UCL, and 50+ research papers and patents. Some of his recent work has been published at conferences including KDD, WWW, SIGIR, RecSys and WSDM. He has co-taught a number of tutorials and summer school courses on the topics of learning from user interactions, marketplaces, and personalization.