National (PID) Graph

The National Graph project built a national-scale graph of persistent identifiers. Its partners worked together to connect Australian research organisations to their research outputs, and the graph showed how research institutions, industry and international partners collaborate.

This page is a record of the completed project. It closed with the release described below, and it has no further work planned.

What the project built

The Australian National Graph was the first effort to build a national-scale graph database of persistent identifiers. Based on the Research Graph schema, it linked more than a thousand research organisations to their research outputs. Its final release, version 1.1.6.3 of 24 April 2024, held:

Australian researchers: 259,111
datasets: 27,566
publications: 1,764,309
grants: 88,234
Australian organisations: 4,406

The technology

The project built its graph database on Neo4j, chosen for its scale, its performance and its ease of use. Neo4j gave the project a native graph database, shaped by the Research Graph schema and tuned for fast graph queries.

A view of the National Graph in the Neo4j browser

Neo4j groups nodes by labels. In the National Graph, the labels stood for metadata sources or data types, which gave the graph a logical structure and allowed targeted searches, such as finding researchers with ORCID profiles.

Summary information about the National Graph database

A three-layer architecture

The National Graph used three layers to compose, optimise and share the graph with the project partners. Keeping the major components apart made each one easier to run.

PID Graph

A data pipeline drew records from the major providers of persistent identifiers, such as Crossref, ORCID and DataCite. Microservices in this layer combined the records into one connected Neo4j graph, using the Research Graph schema together with metadata matching, text mining and entity resolution to connect the identifiers.

Optimisation layer

This layer built a streamlined graph of Australian research output. Its central task was to identify organisation names and link them to their identifiers in ISNI, ROR and Wikidata.

Access layer

The access layer was a distributed network of Neo4j databases through which the project partners reached the National Graph. Its decentralised design let Australian universities run access on their own cloud infrastructure and allocate resources as they needed.

The project released the National Graph data to Australian researchers under a Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International Licence.