Reproduce “Query- and Candidate-Aware Long User Sequence Modeling for Sponsored Product Ranking in Search” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Dongyue Xie, Sergey Kolbin, Manavender Malgireddy, Iurii Shcherbak, Weixing Tang, Arushi Jain, Wanchen Shao, Kurt Zimmer and 2 more
The authors develop a sponsored-product search ranker that retrieves shopper actions using both candidate-product similarity and query-product relevance. The model adds query and candidate context tokens and fuses the candidate into each history token. In Wayfair’s production system, it improves offline ranking metrics and raises orders per impression by 11.5% in a live test, while increasing latency about 12%.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.