RecoGym: A Reinforcement Learning Environment for the problem of Product Recommendation in Online Advertising
David Rohde, Stephen Bonner, Travis Dunlop, Flavian Vasile, Alexandros Karatzoglou
TL;DR
RecoGym introduces an RL-based simulator for product recommendation in online advertising that jointly models organic user behavior and bandit ad interactions. Built as an OpenAI Gym environment, it enables evaluating RL policies without inverse propensity scoring variance by simulating user responses to arbitrary policies. The paper formalizes the problem as a two-mode user process, defines notation and a Markov-chain framework, and provides sanity checks and baseline agents to illustrate performance under varying levels of bandit data and ad-fatigue. The contribution aims to bridge recommender systems and reinforcement learning communities and improve alignment between offline metrics and online performance.
Abstract
Recommender Systems are becoming ubiquitous in many settings and take many forms, from product recommendation in e-commerce stores, to query suggestions in search engines, to friend recommendation in social networks. Current research directions which are largely based upon supervised learning from historical data appear to be showing diminishing returns with a lot of practitioners report a discrepancy between improvements in offline metrics for supervised learning and the online performance of the newly proposed models. One possible reason is that we are using the wrong paradigm: when looking at the long-term cycle of collecting historical performance data, creating a new version of the recommendation model, A/B testing it and then rolling it out. We see that there a lot of commonalities with the reinforcement learning (RL) setup, where the agent observes the environment and acts upon it in order to change its state towards better states (states with higher rewards). To this end we introduce RecoGym, an RL environment for recommendation, which is defined by a model of user traffic patterns on e-commerce and the users response to recommendations on the publisher websites. We believe that this is an important step forward for the field of recommendation systems research, that could open up an avenue of collaboration between the recommender systems and reinforcement learning communities and lead to better alignment between offline and online performance metrics.