Graph Data Model: People and Locations Diagram

Property Graph

A vertex has id, label, properties, and its incoming/outgoing edges
An edge has id, tail, head, label, and properties
It equals two relational tables with indexes on both endpoints

Why a Graph

Many-to-many relationships are everywhere in the data
Queries traverse a variable number of hops: WITHIN*0..
One model holds heterogeneous objects: people, places, events

Query Languages

Cypher: pattern matching on vertices and arrows
SPARQL: triples in RDF; Datalog: rules with recursion
4 lines of Cypher become 31 lines of SQL with WITH RECURSIVE

Limits and Trends

One edge links only two vertices: hypergraphs for more
SQL can store graphs but recursive traversal gets clumsy
GQL became an ISO standard in 2024, based on Cypher