Recently I was shown this curious subhierarchy in SNOMED CT
Beer IS A Ethanol?
My first response was that these things are mixtures. And as such should have a contains or has ingredient type of relationship. Beer contains ethanol.
Which then raises the idea that these things are products. (There is such a thing as non-medicinal products!) But not all mixtures are products.
In fact, almost everything that isn’t a pure chemical is a mixture to some degree. And whilst I don’t think we need to define the composition of all substances. We should be mindful of the difference between subclasses and constituents.
So Milk protein is a subclass; if Milk (substance) means “any substance in milk”. But that’s not likely. Also, see Chocolate milk.
Here is where our definition of a substance in SNOMED CT is helpful.
…we allow substance class distinctions that accept or reject an instance based on whether the instance has a sufficient proportion of constituent(s) to fulfill a role…
There are properties of ethanol that are not inherent in beer. For example, being flammable. So it’s not a subtype.
So what about a high alcohol spirit?
That depends, but looking further up the hierarchy is helpful. We’ve got ancestors Chemical and Organic compound. While valid for ethanol, alcoholic drinks are neither of these things.
Similarly for milk. I’d be inclined to say chocolate milk is a type of milk. It’s got all the same properties, plus milk and sugar! There’s a risk an ancestor could be introduced later that invalidates my claim. But for the most part, I think it’s reasonable.
The other rule of thumb is specific chemicals (compounds and elements) generally shouldn’t have subtypes. With the exception of isomers and isotopes etc.
So should “Products” be subclasses of substances? I think the answer is sometimes…