Overview

Learning to rank studies how to order candidates—such as documents, products, or recommendations—so the most useful items appear first. Approaches include pointwise relevance prediction, pairwise methods such as RankNet and Bradley–Terry models, listwise objectives such as LambdaRank, and neural or LLM-based rerankers; reinforcement learning can also optimize rankings for downstream outcomes. Systems are evaluated with metrics such as NDCG and MRR on retrieval benchmarks and recommendation data, then often tested in online experiments. A central tension is that offline ranking gains may not translate to user or business value, while stronger reranking can add latency and remain vulnerable to bias, position effects, and unstable feedback.

Latest papers

CardsList

306 papers34 papers in the last four weeks, up 13% on the four weeks before. 0.3% of all new papers.

  1. Unsupervised Candidate Ranking for Lexical Substitution via Holistic Sentence Semantics

    Sep 15, 2025Zhongyang Hu, Naijie Gu, Xiangzhi Tao +2Learning to Rank

  2. Dropping Just a Handful of Preferences Can Change Top Large Language Model Rankings

    Aug 16, 2025Jenny Y. Huang, Yunyi Shen, Dennis Wei +1Pairwise Preference LearningLLM Evaluation

  3. Algorithmic Shortlisting in Participatory Budgeting

    Aug 7, 2025Juan Ignacio Zambrano, Clément Contet, Jairo Gudiño-Rosero +3Learning to RankPrivacy-Preserving ML

  4. GeoRanker: Distance-Aware Ranking for Worldwide Image Geolocalization

    May 19, 2025Pengyue Jia, Seongheon Park, Song Gao +2Geospatial Representation LearningLearning to Rank

  5. Pairwise Comparisons without Stochastic Transitivity: Model, Theory and Applications

    Jan 13, 2025Sze Ming Lee, Yunxiao ChenPairwise ComparisonPairwise Preference Evaluation

  6. Personalized Execution Time Optimization for Billion-Scale Scheduled Jobs

    Mar 11, 2022Yang Liu, Juan Wang, Idris Malik +7Learning to Rank