You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Warrick Macmillan edited this page Nov 8, 2021
·
7 revisions
On the Grammar of Proof
by Warrick Macmillan
Abstract
The notion of proof in mathematics has manifold interpretations: with the
distinction between formal and informal proofs representing a significant
spectrum from which to develop and discuss them. One mechanism of understanding
this degree of formality is via translation between natural language proofs
and those one writes via a proof assistant. Grammatical Framework (GF),
a domain-specific functional programming language for specifying grammars
which allows for translation by composing a parser and a pretty printer
(or linearizer), can be used to translate informal and formal proofs.
We contrast how different grammars in GF can be utilized for
this task of proof translation, and see how these grammars enable the possibility
of both natural language generation and autoformalization, but also offer
intractable problems at scale.