Should we stop using distance in our location-based data recommendation models?
summary
Argues straight-line distance often misranks results and shows an API + Elasticsearch plugin that computes travel-time ranks from actual transport networks, handling 100,000 locations in under 100 ms. Demonstrates real deployments: a recruitment customer rebuilds models with the travel-time data, StepStone raises conversions by +10%, and distance-only queries can return up to 50% irrelevant results. Shows PCR-test routing and shift-worker use cases and that the system works in production.
AI-written summary · about
chapters (7)
- 0:00Start
- 1:57Geocoding
- 3:32Bounding Box Area
- 4:17Polygons
- 12:24Isochronic Distance Map
- 12:35Iso Distance
- 23:35Elastic Search Plugin
related talks
28m"Where, When, Why, and How: Directi
25mBetter search relevance using Learning to Rank at mobile.de
16m(Travel) time is money! Costing turns, intersections, and ways
37mSearch and Recommendations: 3 Sides of the Same Coin #bbuzz
20mDataEngBytes 2023 - PER-T1-06 - Kamal Hossain
06mLearning from Hometown and Current City: Cross-city POI Recommendation via Interest Drift and ...
41mDynamic Ranking of Search and Recommendation Results in High-Traffic Systems
14mLeveraging ML to Detect Application HotSpots [@scale, of Course!]