Now that the founding dichotomy is chosen, there are additional steps to achieve in order to build what the database shall exemplify, a typology. But the type policy can’t restrict to this preliminary choice. The construction of a typology requires to obey explicit naming rules and is probably closely bond to the evolution of implied knowledge.
Moreover, the administrator of the base shall decide which type of record can stand as a property in which other type or record, etc. Just the kind of decisions Bertrand Russell made when correcting Gotlob Frege with his type theory. Except that here we avoided such paradoxes at print time only by ensuring the recursive process. You can always be a pre-russellian fregean, Unidatab lets you forget about paradoxes (if any bug doesn’t come to refute this point) by making them inoffensive (from an computational point of view). So these steps in type policy depend on user input and are thus already beyond our concern.
Let’s say for calming, that Unidatab doesn’t require the user to construct a typology in order to enjoy its benefits. Type names and alias names can always be chosen on the fly, without disturbing the rest of the base.