Variational Bayesian Personalized Ranking

Bin Liu, Xiaohong Liu, Qin Luo, Ziqiao Shang, Jielei Chu, Lin Ma, Zhaoyu Li, Fei Teng, Guangtao Zhai, Tianrui Li · 2025

Read Variational Bayesian Personalized Ranking by Bin Liu, Xiaohong Liu, Qin Luo, Ziqiao Shang, Jielei Chu, Lin Ma, Zhaoyu Li, Fei Teng, Guangtao Zhai, Tianrui Li (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2503-11067.txt.

Description

Bin Liu and co-authors reformulate recommendation ranking as variational inference in a 2025 paper. Pairwise learning drives most implicit-feedback recommenders, yet sparse, noisy clicks and popularity bias corrupt it. Popular items get shown more, then clicked more, in a loop.

Their VarBPR treats the hidden index variables behind each comparison probabilistically. Training splits into an inference stage and a learning stage, with closed-form posteriors. A prior sets the target exposure pattern, and a temperature knob controls debiasing strength. The design gives operators principled control over how much popularity should count. Recommenders could serve long-tail taste without abandoning statistical rigor.

Catalog record

CREATORBin Liu, Xiaohong Liu, Qin Luo, Ziqiao Shang, Jielei Chu, Lin Ma, Zhaoyu Li, Fei Teng, Guangtao Zhai, Tianrui Li
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE88 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2503-11067

Browse the full library · About the Tower of Babel