GEON — Geospatial Experience-Oriented Notation

~24%

fewer tokens than GeoJSON

8.8

facts per 100 tokens

1.2x

more places per context window

10+

semantic field categories

Interactive Pipeline

See the full flow

Click a place and watch data flow through the pipeline — from a point on the map to structured GEON, equivalent GeoJSON, and a grounded LLM response.

Map Click markers to switch

GEON

GeoJSON

LLM Response

Comparison

Why GEON?

Traditional geospatial formats are built for machines. GEON is built for understanding.

	GeoJSON	WKT	GEON
Readability	Machine-first	Machine-first	Human & LLM-first
Semantics	Flat properties bag	None	Structured fields
Experience	Absent	Absent	Core feature
Relationships	Requires computation	Absent	Explicit
Temporal	Limited	None	Integrated
Token efficiency	~24% more	Geometry only	Baseline

Format Features

What GEON captures

Every aspect of a place — from its physical form to how it feels to be there.

Identity & Geometry

Name, type, coordinates, boundary polygons, elevation, area — the spatial foundation.

Purpose & Function

What a place is for: commerce, gathering, worship, recreation — structured by category.

Experiential Qualities

Openness, noise, safety, pace, enclosure — 13 scales that describe how a place feels.

Spatial Relationships

Adjacencies, connectivity, containment, viewsheds — how places relate to each other.

Temporal Patterns

Footfall rhythms, opening hours, seasonal events, lifespan — places change over time.

Provenance & Trust

Source attribution, confidence levels, update timestamps — know where your data came from.

Reference

Field overview

Three required fields, plus a rich set of optional semantic, relational, temporal, and provenance fields.

Core Fields

PLACE required — name
TYPE required — classification
LOCATION required — lat, lon
BOUNDARY optional — polygon
EXTENT optional — bounding box
AREA optional — surface area
ELEVATION optional — height
ID optional — identifier

Semantic Fields

PURPOSE — activity types
EXPERIENCE — 13 quality scales
CHARACTER — descriptive traits

Relational Fields

ADJACENCIES — neighbours
CONNECTIVITY — networks
CONTAINS — nested places
PART_OF — parent place
VIEWSHEDS — visible landmarks

Temporal & Provenance

TEMPORAL — time-varying patterns
LIFESPAN — age and permanence
SOURCE — data origins
CONFIDENCE — quality indicators
UPDATED — last modified

Extended Domain

BUILT_FORM ECOLOGY INFRASTRUCTURE
DEMOGRAPHICS ECONOMY VISUAL

Interactive Tools

Try it now

Explore GEON with two browser-based tools — no install required.

Map Explorer

Click the map to fetch OSM features, search by name, or paste GeoJSON. See instant GEON output with syntax highlighting, validation, and token comparison.

Place Builder

Fill out every GEON field with a structured form. Purpose chips, experience dropdowns, dynamic key-value editors — live GEON output as you type.

Official Libraries & SDKs

Build with GEON

Native parsers, generators, and validators for your favorite languages.

# Install from GitHub
pip install "git+https://github.com/jwilliamsresearch/GEON.git#subdirectory=geon-py"

# Usage
import geon

place = geon.parse("""
PLACE: Victoria Park
TYPE: public_space
LOCATION: 52.9403, -1.1340
PURPOSE:
  - recreation
  - ecology
""")

print(place.place)      # "Victoria Park"
print(place.location)   # (52.9403, -1.1340)

# Add git dependency
cargo add --git https://github.com/jwilliamsresearch/GEON.git geon-rs

# Usage
use geon_rs::Geon;

let input = r#"
PLACE: Victoria Park
TYPE: public_space
LOCATION: 52.9403, -1.1340
"#;

let place: Geon = input.parse()?;
println!("Loaded: {}", place.identity.name);

# Install from GitHub (as workspace or submodule)
npm install github:jwilliamsresearch/GEON

# Usage
import { parse } from 'geon-js';

const input = `
PLACE: Victoria Park
TYPE: public_space
LOCATION: 52.9403, -1.1340
`;

const place = parse(input);
console.log(place.identity.name);

A human & AI-readableformat for place.