PermR approximates constrained ILP revenue maximization via neighbor swaps, reaching 63% of optimal gains within latency bounds and delivering 2% revenue lift in a 56M-query online test.
Re-ranking with constraints on diversified exposures for homepage recommender system.arXiv preprint arXiv:2112.07621, 2021
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.IR 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
Fast and Feasible: Permutation-based Constrained Reranking for Revenue Maximization
PermR approximates constrained ILP revenue maximization via neighbor swaps, reaching 63% of optimal gains within latency bounds and delivering 2% revenue lift in a 56M-query online test.