Name
Abstract | ||
|---|---|---|
Online learning to rank is a core problem in information retrieval and machine learning. Many provably efficient algorithms have been recently proposed for this problem in specific click models. The click model is a model of how the user interacts with a list of documents. Though these results are significant, their impact on practice is limited, because all proposed algorithms are designed for specific click models and lack convergence guarantees in other models. In this work, we propose BatchRank, the first online learning to rank algorithm for a broad class of click models. The class encompasses two most fundamental click models, the cascade and position-based models. We derive a gap-dependent upper bound on the $T$-step regret of BatchRank and evaluate it on a range of web search queries. We observe that BatchRank outperforms ranked bandits and is more robust than CascadeKL-UCB, an existing algorithm for the cascade model. |
Year | Venue | Field |
|---|---|---|
2017 | ICML | Convergence (routing),Online learning,Pattern recognition,Ranking,Regret,Upper and lower bounds,Computer science,Cascade,Artificial intelligence,Machine learning,Click model |
DocType | Citations | PageRank |
Conference | 7 | 0.47 |
References | Authors | |
17 | 6 | |
Name | Order | Citations | PageRank |
|---|---|---|---|
| Masrour Zoghi | 1 | 172 | 9.76 |
| Tomás Tunys | 2 | 7 | 0.81 |
| Mohammad Ghavamzadeh | 3 | 814 | 67.73 |
| Branislav Kveton | 4 | 455 | 49.32 |
| Csaba Szepesvari | 5 | 49 | 10.30 |
| Zheng Wen | 6 | 278 | 40.04 |