Following on from my first blog, here are some thoughts around a project I'm currently involved in relating to ontology.
The idea for a New Zealand all-of-government ontology first came up – as far as I am aware – in 2018. At this time Archives New Zealand (Archives NZ) was grappling with how to improve access to its holdings, reform of the State Sector Act was encouraging cross-agency work, and other departments, notably Statistics NZ and the Department of Internal Affairs, were working hard on standards and models to improve information sharing. The stars were aligning, but first we had to be clear about a few things. Like what did we mean by ontology, and come to that, what did we mean by all-of-government? And assuming we can agree on definitions, how is this thing even going to help?
In the last few years the word ‘ontology’ seems to have started occupying the conceptual space that ‘taxonomy’ did in the 1990s and 2000s, in that to some people it simply means a set of vocabularies relating to a specific domain or organisation (see https://www.w3.org/standards/semanticweb/ontology for more on this). As far as I’m concerned, that’s not wrong, it’s just half the story, and it’s the missing half that turns it into a best-seller. Vocabularies, yes, but in an ontology the terms in those vocabularies will belong to designated categories (classes or entities) which support the creation of rich relationships between them. So, as a basic example, we have a list of musical works, and we have a list of artists, and we can relate the works to the artists by creating explicit relationships to show who wrote what, and who performed what. The vocabularies in the ontology can be used as authoritative term sets and a means to interoperability, and the ontology as a whole can be used to support findability, autocategorisation and classification, inferencing, knowledge graphs, and more besides.
As for ‘all-of-government’, that could refer to an ontology which focuses just on the structure of government, or which covers all government activities, or one that is used by all government agencies, or all three. In New Zealand there are already three ontologies which were designed to cover all the activities of their respective agencies (Inland Revenue, Ministry of Justice and the Department of Conservation), plus ontologies developed for specific initiatives in at least another three agencies. Almost all of them use the same classes as their backbone or upper ontology (for example Party, Activity, Event, Authority), so in theory there should be scope for harmonisation and cross-walks (and in most areas I would say that’s true, but there are always gotchas, of course, of which more another time).
So what then is the most useful interpretation of ‘all-of-government’? Which brings us back to: how could ontology help New Zealand government, and therefore benefit New Zealanders?
As I said at the start, Archives NZ needs to provide easy, intuitive access to the digital content it receives (or will receive) from government departments, who of course all have their own way of describing their records. This diversity of description may work for each agency, but is unlikely to be helpful to someone trying to search across the archives. Is there a way to offer standardised terminology for agencies to build on, ie useful lists that all or several agencies might incorporate in their own systems? Such pre-existing lists would save them time and make searching across documents and data so much easier. There’s already a lot of work being done in this area: could that be incorporated into an ontology? Or is it better to try to standardise language as records are received by Archives New Zealand? And how could that happen?
One particular area that has caused increasing concern here is the difficulty of tracing the ‘family tree’ of government: to find out which agency was responsible for what at any given time. This is crucial for individuals trying to establish where records might be kept, whose responsibility a particular outcome was, or what the legal situation was in a given situation. How can you find the records when you don’t even know the name of the relevant department? A solution to this would be an all-of-government ontology which lists all agencies and links them to their predecessors, to their parent or child organisations, to their portfolios, and even to their ministers.
But this is just some of what an all-of-government ontology could be. In order to dig into this more concretely, explore the benefits, and come up with some recommendations, my colleague Liz Wilson and I were charged with writing an options paper for Archives NZ. It was focused on New Zealand needs and activities, but we looked briefly at what other countries are doing and at key international standards. Most importantly, we interviewed people working in NZ government, research and the public sector to gather their views and pain points in relation to information findability, sharing and management, and were impressed time and again by our interviewees’ insights and willingness to share them. The more we talked to them, the more it really did feel that the time is right. And what we mean by ‘ontology’ and ‘all-of-government’ is a lot clearer, as you can see for yourselves (I hope!) in the options paper, which is available here https://archives.govt.nz/publications/all-of-government-ontology-options-paper. The next step now is to put together the development plan. Fun times, as they say here!