Research & intelligence · Global City Intelligence
Global City Intelligence publishes its place-classification methodology
A methodology note explains why name-based classification fails, using a park called Lakeview that contains no lake.
On 26 August 2026, Global City Intelligence published a methodology note describing how it classifies places held in its corpus.
What was published
The note addresses a specific failure mode: classifying a place by what its name appears to say. A park named “Lakeview” is not a lake, and a corpus that infers category from name will record it as one. The project describes classifying places from structured data instead, and documents where that approach itself breaks down.
Why HELPERG is recording it
Published methodology is what makes a dataset checkable. A project that explains how it classifies — including the cases it gets wrong — can be evaluated by someone who does not work on it. That is worth noting as an ecosystem development in its own right, separately from how large the corpus is.
Where to read it
The methodology note is published by Global City Intelligence and linked above. HELPERG does not reproduce it here.