DCM Bandits: Learning to Rank with Multiple Clicks

TL;DR

This paper introduces DCM bandits, an online learning algorithm for ranking web pages based on user click behavior modeled by the dependent click model, optimizing satisfaction probability efficiently.

Contribution

It presents the first practical, regret-optimal algorithm for learning to rank with multiple clicks under the dependent click model, with theoretical guarantees and empirical validation.

Findings

Algorithm performs well on synthetic data.

Algorithm is robust to model misspecification.

Provides tight regret bounds.

Abstract

A search engine recommends to the user a list of web pages. The user examines this list, from the first page to the last, and clicks on all attractive pages until the user is satisfied. This behavior of the user can be described by the dependent click model (DCM). We propose DCM bandits, an online learning variant of the DCM where the goal is to maximize the probability of recommending satisfactory items, such as web pages. The main challenge of our learning problem is that we do not observe which attractive item is satisfactory. We propose a computationally-efficient learning algorithm for solving our problem, dcmKL-UCB; derive gap-dependent upper bounds on its regret under reasonable assumptions; and also prove a matching lower bound up to logarithmic factors. We evaluate our algorithm on synthetic and real-world problems, and show that it performs well even when our model is misspecified. This work presents the first practical and regret-optimal online algorithm for learning to rank with multiple clicks in a cascade-like click model.