Notice: The reproducibility variables underlying each score are classified using an automated LLM-based pipeline, validated against a manually labeled dataset. LLM-based classification introduces uncertainty and potential bias; scores should be interpreted as estimates. Full accuracy metrics and methodology are described in [1].

Online Learning to Rank with Top-k Feedback

Authors: Sougata Chaudhuri, Ambuj Tewari

JMLR 2017 | Venue PDF | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental We empirically demonstrate the performance of our algorithms on simulated and real world data sets.
Researcher Affiliation Academia Department of Statistics 1 Department of Electrical Engineering and Computer Science 2 University of Michigan Ann Arbor, MI 48109, USA
Pseudocode Yes Algorithm 1 Rankingwith Top-k Feedback(RTop-k F)Non Contextual
Open Source Code No The paper does not contain any explicit statement or link indicating the release of source code for the methodology described.
Open Datasets Yes We compared the various ranking functions on two large scale commercial data sets. They were Yahoo s Learning to Rank Challenge data set (Chapelle and Chang, 2011) and a data set published by Russian search engine Yandex (IM-2009). The Yahoo data set had 19944 unique queries with 5 distinct relevance levels, while Yandex had 9126 unique queries with 5 distinct relevance levels.
Dataset Splits No The paper describes an online learning setting, where data is processed sequentially over time (T rounds or iterations), rather than partitioned into explicit, fixed training, validation, and test sets. Therefore, traditional dataset splits are not provided in the context of this online learning problem.
Hardware Specification No No specific hardware details such as CPU, GPU models, or memory specifications are mentioned in the paper.
Software Dependencies No The paper does not provide specific software dependencies with version numbers, such as programming languages, libraries, or solvers.
Experiment Setup Yes All the online algorithms, other than the fully random one, involve learning rate η and exploration parameter γ... In our experiments, for each instance of Algorithm 2, we selected a time varying η = 0.01 t2/3 and γ = 0.1 t1/3 , for round t . We fixed ϵ = 0.01. For List Net, we selected η = 0.01 t1/2 , since regret guarantee in OGD is established with η = O(T 1/2).