Reproduce “LUMIR: Decoupled Multimodal Representation Learning from Heterogeneous Data for Next-POI 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
Zehui Wang, Wolfram Höpken, Dietmar Jannach
LUMIR learns reusable representations of points of interest from metadata, reviews, photos, and user-visit transitions, then supplies them as fixed inputs to sequential recommenders. It combines multimodal compression with sequence alignment without jointly optimizing representations for a particular recommender. On Yelp, the method improves HitRate@10 by 24.2% over the strongest baseline with BERT4Rec, but gains vary across three tested architectures.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.