Snowflake and ontology
Snowflake’s new semantic views are an example. They let you define business metrics, alias columns, and create reusable query logic. All valuable things for data teams.
But sgot nothin to do with ontology: - no class hierarchies or subsumption reasoning - no formal axioms - no open-world assumption - no RDF/OWL-style typing - no inference engine - relationship semantics that go no deeper than a foreign key
What Snowflake (and dbt, and Looker) call a “semantic layer” is really a ‘business logic layer’, a way to make SQL reusable and governed. That’s valuable engineering. But it’s not ontology.
The confusion isn’t trivial. Systems that conflate the two end up with data that’s queryable but not reasoned over. You can answer “what was revenue last quarter” but not “which of these entities are legally the same organization under EU merger rules.”
The distinction matters. Words matter.