Causality, Counterfactuals, Sequential Decision-Making & Reinforcement Learning for Recommender Systems

Recommender systems are increasingly modelled as repeated decision making processes that decide which items to recommend to a given user. Each decision to recommend an item or slate of items has a significant impact on immediate and future user responses, long-term satisfaction or engagement with the system, and possibly valuable exposure for the item provider. This interactive and interventionist view of recommendation uncovers a plethora of unanswered research questions, as it complicates the classical offline evaluation and learning procedures in the field.
A first challenge is to develop a deep understanding of causal inference to reason about (possibly unintended) consequences of the recommender, and a notion of counterfactuals to answer common “what if”- type questions in learning and evaluation. Advances in the intersection of these fields can foster progress in effective, efficient and fair learning and evaluation from logged data. This will be the focus of the CONSEQUENCES workshop.
A second challenge is to optimise a multi-step decision-making process, where a stream of interactions occurs between the user and the system. Deriving reward signals from these interactions, and creating a scalable, performant, and maintainable recommendation model to use for inference is a key challenge for machine learning teams, both in industry and academia. To make the system design a bit more tractable, these environment interactions are often viewed as independent; but to further improve and scale recommender systems, the models must take into account the delayed effects of each recommendation, and begin reasoning/planning for longer-term user satisfaction, leveraging techniques such as Reinforcement Learning (RL). This will be the focus of the REVEAL workshop.

These topics have been emerging in the Recommender Systems community for a while, our workshop aims to bring a dedicated forum to learn and exchange ideas.
To this end, we welcome contributions from both academia and industry and bring together a growing community of researchers and practitioners interested in sequential decision making, reinforcement learning, offline and off-policy evaluation, batch policy learning, fairness in online platforms, as well as other related tasks, such as A/B testing.

Organizers

CONSEQUENCES

• Olivier Jeunen, Amazon
• Thorsten Joachims, Information Science and Computer Science, Cornell University
• Yuta Saito, Department of Computer Science, Cornell University
• Harrie Oosterhuis, Radboud University and Twitter
• Flavian Vasile, Criteo

REVEAL

• Paige Bailey, Anyscale
• Maria Dimakopoulou, Spotify
• Ying Li, Netflix
• Richard Liaw, Anyscale
• Yves Raimond, Netflix

Website

https://sites.google.com/view/consequences2022/home

Date

Two full days