No, sorry, that is not what I meant. When I said "current data model" I meant the "currently proposed data model". Sorry for being sloppy.
So it assumes new entity types for Lexemes, which are not just special forms of Items. And it is not reliant on an underlying graph model.
Daniel already sketched out how 'product' may look like. Since the current implementation does not support Lexemes, I can not just put it on Labs.
But there could be a Lexeme "produc-" which is pointed to from Daniel's Lexemes for "to produce", "production", "producer", etc., which could point to "produc-" via a statement, say, "root word" or similar. In the end, it really is unclear whether that is correct or not, but it sure is a possibility that can represented with the currently proposed data model. Which properties exist, how they are linked to each other, etc., is all up to the collaborative decisions which the community has to make.