Principled and Scalable Diversity-Aware Retrieval via Cardinality-Constrained Binary Quadratic Programming

Qiheng Lu & Nicholas D. Sidiropoulos · 2026

Read Principled and Scalable Diversity-Aware Retrieval via Cardinality-Constrained Binary Quadratic Programming by Qiheng Lu & Nicholas D. Sidiropoulos (2026) 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-2604-02554.txt.

Description

Lu and Sidiropoulos tackle picking search hits that are both apt and varied, a need when AI answers draw on retrieved text. They frame the choice as a binary quadratic program with a cap on how many passages are picked, trading aptness for variety with one clear knob. They build a tight continuous relaxation and a Frank-Wolfe method with proof of convergence. Tests show it beats rivals on the aptness-variety frontier while running much faster.

Catalog record

CREATORQiheng Lu & Nicholas D. Sidiropoulos
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE50564
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2604.02554

Browse the full library · About the Tower of Babel