The ab urbe condita construction is an NP consisting of a head nominal H and an attribute A such that the designata of H and A form an argument and a predicate, respectively, which form a proposition which is the meaning of the NP. In other words, the meaning of the NP is not (as usual) “(kind of) H which is A”, but instead “the fact/event/circumstance that H is A”. The construction is named after a Latin example of it:

Structurally, ab urbe condita is a prepositional phrase with a preposition governing a noun phrase in the ablative, whose head is the noun urbe, which in turn has an attribute which is the participle condita. However, years are not counted from the city but from the event of its foundation. Semantically, therefore, urbe condita is a manifestation of a reified proposition, something that would more commonly be represented by a nominalization like urbis conditio “foundation of the city”.

An English example of the construction is the expression in B2, taken from this website.

Here, too, one is not talking about a consequence of the overlap, but of its incompleteness. Similarly, when I say the weak support has made me doubt, it is not the support that has made me doubt but rather its weakness. Thus, the attribute appearing in the construction need not even be deverbal; a simple adjective may do. In all these cases, attribution is the structural manifestation of a predication, and the roles of head and dependent in the semantic representation are countericonically reversed in the structural representation.

Reference

Bolkestein, A. Machtelt 1981, "Factivity as a condition for an optional expression rule in Latin: the "ab urbe condita" construction and its underlying representation." Bolkestein, A. Machtelt et al., Predication and expression in Functional Grammar. London etc.: Academic Press; 205-233.