Online Learning to Rank in Stochastic Click Models
Authors: Masrour Zoghi, Tomas Tunys, Mohammad Ghavamzadeh, Branislav Kveton, Csaba Szepesvari, Zheng Wen
ICML 2017 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | We derive a gap-dependent upper bound on the T-step regret of Batch Rank and evaluate it on a range of web search queries. We observe that Batch Rank outperforms ranked bandits and is more robust than Cascade KL-UCB, an existing algorithm for the cascade model. |
| Researcher Affiliation | Collaboration | 1Independent Researcher, Vancouver, BC, Canada (Part of this work was done during an internship at Adobe Research) 2Czech Technical University, Prague, Czech Republic 3Deep Mind, Mountain View, CA, USA (This work was done while the author was at Adobe Research) 4Adobe Research, San Jose, CA, USA 5University of Alberta, Edmonton, AB, Canada. |
| Pseudocode | Yes | Algorithm 1 Batch Rank, Algorithm 2 Display Batch, Algorithm 3 Collect Clicks, Algorithm 4 Update Batch |
| Open Source Code | No | The paper does not provide a direct link or explicit statement about the availability of the source code for the Batch Rank algorithm or other methods described in the paper. It only mentions 'Py Click (Chuklin et al., 2015), which is an open-source library of click models for web search.' |
| Open Datasets | Yes | We experiment with the Yandex dataset (Yandex), a dataset of 35 million (M) search sessions, each of which may contain multiple search queries. |
| Dataset Splits | No | The paper mentions using the Yandex dataset for experiments and averaging regret over periods, but it does not specify explicit training, validation, and test splits (e.g., 70/15/15 percentages or specific sample counts for each split). |
| Hardware Specification | No | The paper does not provide specific hardware details (e.g., GPU/CPU models, memory) used for running the experiments. |
| Software Dependencies | No | The paper mentions 'Py Click (Chuklin et al., 2015), which is an open-source library of click models for web search' but does not specify any version numbers for this or other software components. |
| Experiment Setup | Yes | In each query, our goal it to rerank L = 10 most attractive items with the objective of maximizing the expected number of clicks at the first K = 5 positions... The reported regret is averaged over periods of 100k steps to reduce randomness. |