Research map / Bandits and online learning / Contextual bandits and Thompson sampling
DCM Bandits: Learning to Rank with Multiple Clicks
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| Use all observed clicks (learn from every observed attraction weight) instead of only first click (ablation) | dcmKL-UCB · B_LB(L=16, A*=[4], p=0.2, Delta=0.15, gamma=0.5) | cumulative regret: dcmKL-UCB regret clearly lowest vs First-Click | Helped |
| Use all observed clicks instead of only last click (ablation) | dcmKL-UCB · B_LB(L=16, A*=[4], p=0.2, Delta=0.15, gamma=0.5) | cumulative regret: dcmKL-UCB regret clearly lowest vs Last-Click | Helped |
| Replace DCM-structured learning with ranked bandit (KL-UCB per position) (ablation) | dcmKL-UCB · B_LB(L=16, A*=[4], p=0.2, Delta=0.15, gamma=0.5) | cumulative regret: RankedKL-UCB regret about three times higher | Hurt |
| Use dcmKL-UCB instead of ranked bandits (RankedKL-UCB, RankedExp3) | Yandex real-world queries · Yandex (20 queries) | average regret at n=10k: at least two times lower than best baseline | Helped |
| Increase common termination probability gamma (fewer clicks) (ablation) | dcmKL-UCB · B_LB(L=16, p=0.2, Delta=0.15), K varied | n=10^5 regret | Hurt |
| Increase number of recommended items K (ablation) | dcmKL-UCB · B_LB(L, A*=[K], p=0.2, Delta, gamma=0.8) | n=10^5 regret | Helped |
Closest papers by meaning
- Online Learning to Rank in Stochastic Click Models (ICML 2017 · similarity 0.90)
- UniRank: Unimodal Bandit Algorithms for Online Ranking (ICML 2022 · similarity 0.89)
- Bandits Meet Mechanism Design to Combat Clickbait in Online Recommendation (ICLR 2024 · similarity 0.88)
- Adaptively Learning to Select-Rank in Online Platforms (ICML 2024 · similarity 0.86)
- Leveraging Offline Data in Linear Latent Contextual Bandits (ICML 2025 · similarity 0.85)
- PageRank Bandits for Link Prediction (NeurIPS 2024 · similarity 0.85)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.