Google Research Introduces ME-POIs: A Mobility-Informed Framework that Adds “How a Place Is Used” to Text-Based POI Embeddings

Google Research introduced ME-POIs, a mobility-informed framework that adds human movement patterns to text-based point-of-interest (POI) embeddings. In evaluations on Los Angeles and Houston mobility data, ME-POIs improved 34 of 35 model-task pairings in Los Angeles, achieving up to 81.9% relative F1 improvement on visit intent and a 24.7% reduction in mean absolute error on busyness prediction. A mobility-only variant outperformed Gemini text embeddings on price-level classification.

framework that folds aggregate human movement into text-based place embeddings. Language models describe what a place is; they miss how it is used. ME-POIs encodes each visit as a contextualized vector and aligns it with one learnable prototype per POI through contrastive learning, then transfers visit distributions from data-rich anchors to the long tail across three spatial scales. Across five map-enrichment tasks on Los Angeles and Houston mobility data, adding ME-POIs improved 34 of 35 model-task pairings in Los Angeles — up to 81.9% relative F1 on visit intent and a 24.7% MAE reduction on busyness. A mobility-only variant beat Gemini text embeddings on price-level classification. The post Google Research Introduces ME-POIs: A Mobility-Informed Framework that Adds “How a Place Is Used” to Text-Based POI Embeddings https://www.marktechpost.com/2026/08/24/google-research-introduces-me-pois-a-mobility-informed-framework-that-adds-how-a-place-is-used-to-text-based-poi-embeddings/ appeared first on MarkTechPost https://www.marktechpost.com .