Semantic interoperability has an operational side that is easy to forget, even though it is in the name. It needs a source system and a target system, and a business reason for them to exchange data by its meaning. Only then can each carry on with its own internal logic, one providing the data and the other consuming it.
Two contracts
Digital service provision already runs on contracts. They define the conditions under which a service can be accessed.
Semantic interoperability needs a second contract: a contract on the data, or more precisely a metadata contract. It states what the data means in relation to the internal business logic on both sides.
Like the service contract, the metadata contract must use a formalism that is not open to interpretation. It must also allow strict logical entailment for anything that is not stated explicitly. So what do we have that describes the meaning of data with the properties of a contract? Ontologies, and in particular those based on description logics.
The benefit, and the drawback
Ontologies can be composed from existing parts. Over time, reuse turns into convergence of descriptions, and convergence of descriptions is increased semantic interoperability.
Their drawback is not technical. It is the common practice of not practising reuse.
The history of attempts is long: monolithic efforts, foundational ontologies, upper-level ontologies, collections of design patterns. After twenty years we should see a converging effect across public-sector models, much as application profiles have multiplied around shared catalogue vocabularies. We largely don’t.
Enabling and tracking convergence
In my work with the European Commission’s semantic interoperability action, SEMIC, I have tried to do two things: enable convergence, and track it. Where there is divergence, consolidate it.
That is why I initiated the Semantic Registry: a way to coordinate the design decisions that, taken consistently and coherently, increase semantic interoperability across administrations.
One cannot design semantic interoperability directly. One can create the conditions for it to emerge, which I believe was the intention of the web’s architecture all along.
If you work for a national administration and want to converge towards European semantic interoperability, start a conversation with us.