Learning to Rank For Push Notifications Using Pairwise Expected Regret

TL;DR

This paper introduces a novel pairwise ranking loss based on expected regret to improve personalized push notification ranking, demonstrating superior performance in simulations and real-world social network experiments.

Contribution

The paper proposes a new ranking loss function that incorporates expected regret, addressing unique challenges in mobile push notification ranking.

Findings

Outperforms prior methods in simulated environments.

Achieves better engagement metrics in a production social network setting.

Validates effectiveness of the regret-based loss in real-world applications.

Abstract

Listwise ranking losses have been widely studied in recommender systems. However, new paradigms of content consumption present new challenges for ranking methods. In this work we contribute an analysis of learning to rank for personalized mobile push notifications and discuss the unique challenges this presents compared to traditional ranking problems. To address these challenges, we introduce a novel ranking loss based on weighting the pairwise loss between candidates by the expected regret incurred for misordering the pair. We demonstrate that the proposed method can outperform prior methods both in a simulated environment and in a production experiment on a major social network.