Te pūrongo kōwhiringa raupapa mahi ā-kawanatanga katoa

All-of-government ontology options paper

This options paper for an all-of-government ontology has been commissioned by Archives New Zealand, Te Rua Mahara o te Kawanatanga. Archives New Zealand is the agency that holds and provides access to government information of archival value. It establishes the regulatory framework for information and records management across the public sector, and the Public Records Act 2005 provides for the Chief Archivist to exercise a leadership role in this, setting standards for the creation, maintenance and efficient management of public records (including data). The purpose of this options paper is to explore the benefits of an all-of-government ontology in helping agencies to find, use, manage and share their information and data.

Ontologies provide common vocabularies, concept definitions and relationships between concepts. They support interoperability, the semantic analysis of content, autocategorisation of content and automation of business processes, knowledge representation and the use of artificial intelligence and machine-learning tools, widely thought to be the future of information management.

In developing the paper, we have explored options, benefits, demand, feasibility and practical implementation ideas for such an ontology with a wide range of stakeholders in the public, private and academic sectors. We appreciate the wide-ranging support from such a diverse community, particularly The Digital Public Service (Government Chief Digital Officer) and Statistics New Zealand (Government Chief Data Steward).

From an Archives business perspective, an all-of-government ontology will enable consistent categorisation of government information, and greatly improve access to holdings from multiple different points of view, both cultural and multilingual. But the benefits to such a semantic tool are arguably much wider across government, providing the ability to successfully evolve from our current siloed information and data architectures. An all-of-government ontology could support data sharing, data lineage, automatic categorisation and classification of content, and data analytics. There is already existing work, tools, standards and thinking in New Zealand which can be built upon to take this forward, and great enthusiasm expressed by interviewees: this is an exciting opportunity to look at how modern techniques and technologies can move Archives New Zealand forward in its aims, while also providing substantial benefit across government.

An all-of-government initiative such as an ontology necessarily requires different parts of the government information ecosystem to work together, and so we have explored options that propose joint efforts from key agencies for the initial stages, progressively including sector domains on a modular basis, but which may also include working with third parties. The following are our options for an all-of-government ontology. Some can stand independently, some rely on other options:

1. Repository: to create hub or repository for government ontologies and taxonomies as a means to providing standards for terminology;

2. Foundation Ontology: to design and build a Foundation Ontology covering the key entities relevant to government to provide authoritative concepts (with URIs and definitions, and to include Māori concepts and te reo), and the fundamental building blocks for domain-specific ontologies;

3. Government organisational lineage: within the Foundational Ontology design, to prioritise the development of the classes, relationships and properties required to track the names and remits of Government departments through time;

4. Domain ontologies: to also prioritise the development of classes, relationships and properties for two specific domains - health and housing, since both are ongoing reorganisations, and senior leaders have expressed strong support for this work;

5. Autocategorisation and classification: to build on the Foundation Ontology to create terms, term properties and relationships which support autocategorisation and autoclassification of records when combined with the appropriate rulebase and algorithms;

6. Ontology service: to offer a service which incorporates the guardianship and publication of authoritative ontology models and associated taxonomies, and the provision of a suite of utilities for use in creating, using and disseminating them;

7. Third party platforms: to explore the potential of third party platforms such as the Azure cloud and associated technology stack (e.g. Syntex and Cortex, advanced artificial intelligence tools in M365) incorporating the Foundation Ontology (and potentially domain ontologies) to support processes for autocategorisation and the application of archival business processes.

From these options our recommendations are:

• To create a Foundation Ontology covering the key entities and relationships relevant to New Zealand government;

• To work alongside the development of the Foundation Ontology to create a Government organisational lineage;

• Within the developing Foundational Ontology work, to prioritise domain ontologies in health and housing to support the Ministry of Health and Kāinga Ora;

• To explore the potential for working with a third party such as Microsoft to integrate an all-of-government ontology into advanced artificial intelligence tools to support the information and data lifecycle.

There is also, in due course, the possibility of offering a full ontology service to agencies, but this is not part of the current recommendations.

Development of an all-of-government ontology will necessitate the creation of a centre of excellence, potentially within the Digital Public Service, which is responsible for building and managing the ontology and liaising with agencies.

Read the full All of government options paper (PDF, 2 MB)

Contact us for an accessible version of this PDF.