AI TrackerFrontier desk · AEST
Back to news
normalaitech

SemantEq: Faithful Bidirectional Equivalence of Formal Statements

arXiv cs.AI · 24 Oct 2026, 03:20 AEST

SemantEq introduces a new method for assessing the equivalence of formal statements in proof assistants like Lean 4. This approach shifts the focus from the proof to the statement itself, ensuring a more principled evaluation of logical equivalence.

Key facts

Read aloud

SemantEq introduces a new method for assessing the equivalence of formal statements in proof assistants like Lean four. This approach shifts the focus from the proof to the statement itself, ensuring a more principled evaluation of logical equivalence. Statement autoformalization translates mathematics from natural language into a proof assistant such as Lean four, and models now do it at scale. Given a trusted reference formalization, the natural test is logical equivalence, and it proves too much. SemantEq, a more principled relation, requires each statement to turn into the other only through a chain of allowed steps, such as logical rearrangement and rewriting with the library's own equations. SemantEq-Eval reaches one hundred percent precision and sixty five point six percent recall on the ProofNetVerif test split under adjudicated labels. Separating the generation of equivalence proofs from their acceptance lets a stronger prover raise recall without changing what an acceptance means.

Sources in this cluster