Reproduce “Neither Black nor White: Balancing Semantic and Collaborative Signals with Graph-Informed Semantic IDs (GrIS)” 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
Aleksei Medvedev, Alejandro Ariza-Casabona, Steven Derby, Gonzalo Fiz Pontiveros, Xinyang Shao, Florian Spiess
The authors introduce Graph-Informed Semantic IDs (GrIS), framing item-code construction for generative recommendation as hierarchical graph partitioning. It combines item-description features with behavioral edges, making graph construction and partitioning explicit choices; an empty graph recovers content-only methods. On Amazon Toys, RecDMoN reaches H@10 of 8.16%, versus 5.37% for LETTER, a 52.0% relative gain, within a fixed-backbone evaluation.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.