Reproduce “POI Alias Discovery in Delivery Addresses using User Locations” 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
Tianfu He, Guochun Chen, Chuishi Meng, Huajun He, Zheyi Pan, Yexin Li, Sijie Ruan, Huimin Ren and 6 more
Researchers from Harbin Institute of Technology and JD Intelligent Cities Research introduced a framework to discover aliases for Places of Interest (POIs) in delivery addresses by leveraging user GPS location data. Their approach, which uses aggregated user mobility profiles, achieved F1-scores up to 0.621 in single-city evaluation, significantly outperforming traditional text-based alias discovery methods.
Key insights
- Users associated with the same physical POI, regardless of the name used in delivery addresses, exhibit highly similar and spatially clustered GPS location patterns.
- Aggregated user mobility profiles provide a robust, data-driven signal for inferring POI alias relationships that are not discoverable through textual or semantic analysis alone.
- The "crowd wisdom" inherent in ubiquitous user location data offers a scalable and automated alternative to costly and inefficient manual alias dictionary compilation.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.