A Pareto-front-conditioned single recommender model serves multiple online test groups; the paper reports significant offline-to-online alignments, but the significance analysis treats a five-group covariate as though it varied across millions of users.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.IR 1years
2025 1verdicts
REJECT 1representative citing papers
citing papers explorer
-
Identifying Offline Metrics that Predict Online Impact: A Pragmatic Strategy for Real-World Recommender Systems
A Pareto-front-conditioned single recommender model serves multiple online test groups; the paper reports significant offline-to-online alignments, but the significance analysis treats a five-group covariate as though it varied across millions of users.