The knowledge graph

Brasil Data Commons stores everything as one knowledge graph. Instead of one schema per dataset, every fact is a node or an edge in a shared graph, which is why a single API can answer questions across census, health, and economic data at once.

DCIDs

Every node has a DCID (Data Commons Identifier) — a stable, unique string that names it.

• A place: country/BRA, geoId/3550308 (São Paulo municipality)
• A statistical variable: Count_Person, Median_Income_Person
• A property: containedInPlace, typeOf

You use DCIDs everywhere you refer to a thing. When you don’t know one, search for it.

The four things you’ll work with

Concept	What it is	Example DCID
Entity	Any node — a place, an organization, a dataset	country/BRA
Place	A geographic entity (country, state, municipality)	geoId/3550308
Statistical variable	A measurable quantity	Count_Person
Observation	A value of a variable, for an entity, at a date, from a source	(a graph edge)

Observations and provenance

An observation ties four things together: a variable, an entity, a date, and a value — plus the provenance (which dataset and source it came from). Two sources may report the same variable for the same place; Data Commons keeps both as separate facets so you can choose or compare them rather than silently picking one.

Count_Person @ country/BRA @ 2022 = 203_080_756   (source: IBGE Census)

Places are nested

Places link to each other with containedInPlace, forming a hierarchy:

country/BRA
  └─ State (e.g. São Paulo)
       └─ Municipality (e.g. São Paulo city)

This nesting is what lets you ask “all states in Brazil” with a single relation expression — see Query observations.

Where to go next

• Query observations guide — put these concepts to work.
• API reference — the endpoints that expose the graph.