Reproduce “ID Embedding as Subtle Features of Content and Structure for Multimodal 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
Yuting Liu, Enneng Yang, Yizhou Dang, Guibing Guo, Qiang Liu, Yuliang Liang, Linying Jiang, Xingwei Wang
IDSF uses learned item ID embeddings in separate modality-specific paths to enrich content representations and propagate signals through user-item interaction graphs. It combines these representations for recommendation. Across three Amazon datasets, it leads reported metrics for Baby and Sports, but not every Clothing metric. On Baby, Recall@20 is 9.505 versus MICRO’s 8.884, a reported 6.9% improvement; this is benchmark evidence, not universal proof.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.