Reproduce “BRIDGE: Bridging the Semantic Quantization Gap for Generative Recommendation” 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
Yatong Xiao, Mingjie Sun, Peng Wang, Wenjin Wu, Peng Jiang
BRIDGE improves next-item recommendation by generating hierarchical Semantic ID codes with semantic guidance and difficulty-aware refinement. An RQ-VAE creates the codes; BRIDGE anchors a continuous user-history representation, uses neighbor-informed teacher-student training, and allocates more refinement to earlier codes. On Amazon Beauty, it raises Recall@10 to 0.0835 from GR4AD’s 0.0742; gains apply to benchmarks, and ablations do not isolate EAR’s schedule.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.