Online A/B testing evaluates the impact of a new technology by running it in a real production environment and testing its performance on a subset of the users of the platform. It is a well-known practice to run a preliminary offline evaluation on historical data to iterate faster on new ideas, and to detect poor policies in order to avoid losing money or breaking the system. For such offline evaluations, we are interested in methods that can compute offline an estimate of the potential uplift of performance generated by a new technology. Offline performance can be measured using estimators known as counterfactual or off-policy estimators. Traditional counterfactual estimators, such as capped importance sampling or normalised importance sampling, exhibit unsatisfying bias-variance compromises when experimenting on personalized product recommendation systems. To overcome this issue, we model the bias incurred by these estimators rather than bound it in the worst case, which leads us to propose a new counterfactual estimator. We provide a benchmark of the different estimators showing their correlation with business metrics observed by running online A/B tests on a large-scale commercial recommender system. CCS CONCEPTS• Computing methodologies → Learning from implicit feedback; • Information systems → Evaluation of retrieval results;

No abstract

A large portion of online advertising displays are sold through an auction mechanism called Real Time Bidding (RTB). Each auction corresponds to a display opportunity, for which the competing advertisers need to precisely estimate the economical value in order to bid accordingly. This estimate is typically taken as the advertiser's payoff for the target event -such as a purchase on the merchant website attributed to this display -times this event estimated probability. However, this greedy approach is too naive when several displays are shown to the same user. The purpose of the present paper is to discuss how such an estimation should be made when a user has already been shown one or more displays. Intuitively, while a user is more likely to make a purchase if the number of displays increases, the marginal effect of each display is expected to be decreasing. In this work, we first frame this bidding problem with repeated user interactions by using causal models to value each display individually. Then, based on this approach, we introduce a simple rule to improve the value estimate. This change shows both interesting qualitative properties that follow our previous intuition as well as quantitative improvements on a public data set and online in a production environment.

No abstract

Personalised interactive systems such as recommender systems require selecting relevant items from massive catalogs dependent on context. Reward-driven offline optimisation of these systems can be achieved by a relaxation of the discrete problem resulting in policy learning or REINFORCE style learning algorithms. Unfortunately, this relaxation step requires computing a sum over the entire catalogue making the complexity of the evaluation of the gradient (and hence each stochastic gradient descent iterations) linear in the catalogue size. This calculation is untenable in many real world examples such as large catalogue recommender systems, severely limiting the usefulness of this method in practice. In this paper, we derive an approximation of these policy learning algorithms that scale logarithmically with the catalogue size. Our contribution is based upon combining three novel ideas: a new Monte Carlo estimate of the gradient of a policy, the self normalised importance sampling estimator and the use of fast maximum inner product search at training time. Extensive experiments show that our algorithm is an order of magnitude faster than naive approaches yet produces equally good policies.