Disambiguating Explicit and Implicit Geographic References...
summary
Builds language models over a KD-tree adaptive grid and uses hierarchical logistic regression to score cells, plus simple smoothing and beam search for fast queries. Reports Twitter geolocation: 48% within 100 miles in the U.S., 30% worldwide; Flickr text-only hits 66% within 100 miles; Wikipedia geolocation reaches ~90% accuracy. For toponym resolution, creates per-name classifiers from Wikipedia weak labels, compares to a spatial minimality baseline and a GI-statistic word-profile method that achieves up to ~92% on a news dataset.
AI-written summary · about
chapters (6)
- 0:00Start
- 6:51Language modeling approach
- 9:55Hierarchical geo-location with logistic regression
- 11:11Performance: hierarchical logistic regression with kd-tree grid
- 13:45Hierarchical logistic regression beats flat naive Bayes
- 18:45Toponym classifiers Strategy: build a textual classifier per toponym by obtaining indirectly labeled examples from Wikipedia.
related talks
19mThe Effect of Population and "Structural" Biases on Social Media-based Algorithms - A Case Study ...
07mMapping Urban Deprivation
30mLocation Intelligence and Spatial Data Science: from geographic coordinates to business insights
41mLarge Scale Field Delineation
31mWi‑Fi/GSM Geolocation in Large Cities Using ML for 30 Million Users
31mGeocoding deep dive: Tips & power features
06mThe Crowd Wisdom for Location Privacy of Crowdsensing Photos: Spear or Shield?
01h 36mFrom Maps to Models: An End-to-End Journey in Geospatial Machine Learning with Python