Publishing a vocabulary feels like an ending. After months of workshops, drafts and review rounds, the model has a stable address, a version number and a page in the catalogue. The project closes, the deliverable is ticked, and the vocabulary joins the organisation’s list of interoperability assets.
I would like to suggest that publication is the moment a vocabulary becomes a commitment, and that it becomes an asset later, if at all, depending on what other people do with it.
Four words we use as one
Listen to any interoperability meeting and you will hear four words used almost interchangeably. “We have a vocabulary for that.” “The model is published.” “It is one of our semantic assets.” “Upload the artifact to the catalogue.” Very often all four refer to the same file. It is an understandable habit, and it is a curious one in a field whose whole craft consists of asking other people to define their terms.
The cost of the habit is expressive power. When four words collapse into one, we lose the ability to say the things that matter most about interoperability. Kept apart, each word does a precise job.
A semantic model is the way a party conceives its domain: the concepts it distinguishes, how they relate and what it commits to about them. Every organisation has one, whether it lives in an ontology, a database schema or the heads of its experts.
A vocabulary is a published set of terms with their definitions, offered for others to reuse. It is one particular form a model can take when it is shared.
A semantic artifact is the published, identifiable object that carries a model: it has an address and versions, it can be read by people and machines, and someone stands behind it.
A semantic asset is a status an artifact earns, within a particular ecosystem, when the parties there build on it.
With the four words in place, sentences become possible that were impossible before. “Our semantic model of an employment period has drifted from the artifact we claim to reuse.” “This vocabulary is an asset for pension exchanges and a liability for tax reporting.” “We publish fourteen artifacts, and four of them are assets.” Each of those sentences carries a diagnosis and points to a decision. With one blurred word in place of four, every one of them dissolves into “we have a model for that”, and the conversation stops just where it should begin.
What publication creates
EVIDENT, the framework we are developing at METALEVELS, draws its firmest line between the last two of those words, and the line runs exactly through the moment of publication.
On the day it is published, a vocabulary is an artifact. It is explicit, with a stable identifier; interpretable, with definitions a modeller can judge; auditable, with its versions and authorship on record; and formally grounded, so that machines can compute what relates to what. All of that is valuable work, and none of it yet creates value inside an ecosystem. What publication does create is a set of obligations. Someone has to answer questions about the vocabulary, fix its errors, decide how it evolves, keep its versions coherent and tell users when something changes. Those obligations start immediately and last as long as the vocabulary is in use, or until someone has the courage to retire it.
In accounting terms, a newly published vocabulary looks more like a liability than an asset. In EVIDENT’s terms, it becomes an asset when the evidence shows that the parties of an ecosystem use it as a reliable base to build on, for the scope they need to share. The value of a shared vocabulary lies in the exchanges it makes cheaper, and an exchange needs at least two parties who chose the same model.
Asset status is relational
This is the part that tends to surprise people. Whether a vocabulary is an asset depends less on its quality than on its position among a particular set of parties, for a particular scope. The same vocabulary can be an asset in one ecosystem and a liability in another.
Think of schema.org, adopted across a large part of the web to describe products, events and organisations. For search engines and the sites that want to be found by them, it is a remarkable asset: widely used, well understood, cheap to adopt. Bring the same vocabulary into a closed ecosystem of pension administrations that need to agree on what an employment period is, and its value depends entirely on whether those administrations use it for that purpose. If they have converged elsewhere, adopting it adds one more model to map, and the asset turns into a cost.
The reverse happens too. A modest vocabulary, maintained by a small working group and rarely mentioned outside its circle, can be the most valuable thing in an ecosystem, because the parties who need to exchange data have quietly built on it and run it in production.
What the evidence shows
A list of published vocabularies records artifacts. To see assets, EVIDENT looks at what the parties of an ecosystem have done with each vocabulary, and it reads three kinds of evidence from the models they themselves publish. Formal commitment shows which parties import the vocabulary and build on it as a declared dependency. Alignment shows which parties map their own terms to its terms. Adoption shows which parties run it in production, with live data.
The evidence is reported party by party, for the scope in question, rather than folded into a single score. A programme owner can see, for example, that four of six administrations build on a vocabulary and two only map to it, and which two. Next to that evidence sits the vocabulary’s governance record: who maintains it, how it is versioned, which body stands behind it. That record explains the vocabulary’s standing; the behaviour of the parties shows whether it has become an asset.
This matters most for the next modeller. Faced with three candidate vocabularies that all look authoritative on paper, they would otherwise choose on reputation or familiarity. With the evidence in front of them, they can see which choice brings their system closer to the peers they need to exchange with, and whether their decision will strengthen an existing asset or add a new liability.
A simple asset test
If you own or fund vocabularies, these questions will tell you which of them are working as assets today.
- For which parties, and which exchanges, was this vocabulary published? If nobody can say, start there.
- Which of those parties import it and build on it?
- Which of them map their own terms to it, and which run it in production with live data?
- What does it cost to maintain each year, and which exchanges does that cost buy?
- Is there another vocabulary in the same scope that the parties have actually converged on? If so, consider aligning with it, or retiring yours.
Managing vocabularies as a portfolio
A vocabulary programme that asks these questions starts to manage its models as a portfolio. Some vocabularies are clearly assets and deserve investment in stewardship and new versions. Some are promising and need help reaching the parties they were built for. Some have been overtaken, and the most generous thing to do is to point their users towards the model their peers already share. Seeing that portfolio clearly, through the behaviour of the parties around each vocabulary, is what the framework was built to do.
If you would like to look at your own vocabularies this way, start a conversation with us.