Reproduce “Context-based Fast Recommendation Strategy for Long User Behavior Sequence in Meituan Waimai” 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
Zhichao Feng, Junjiie Xie, Kaiyuan Li, Yu Qin, Pengfei Wang, Qianzhong Li, Bin Yin, Xiang Li and 2 more
CoFARS, developed by Beijing University of Posts and Telecommunications and Meituan, is a two-stage recommendation strategy designed for food delivery platforms that efficiently processes long user behavior sequences by leveraging context similarity, achieving a 4.6% increase in CTR and a 4.2% increase in GMV.
Key insights
- User preferences tend to be consistent within similar contexts, enabling the efficient selection of relevant sub-sequences from long behavior histories.
- Jensen-Shannon divergence offers a precise way to measure context similarity by directly comparing the probability distributions of user preferences.
- A prototype-based approach effectively reduces computational complexity by clustering similar contexts and retrieving only the most relevant historical behaviors.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.