Explain: A Tool for Performing Abductive Inference

This paper describes a tool called Explain for performing abductive inference. Logical abduction is the problem of finding a simple explanatory hypothesis that explains observed facts. Specifically, given a set of premises Γ and a desired conclusion φ, abductive inference finds a simple explanation ψ such that Γ ∧ ψ |= φ, and ψ is consistent with known premises Γ . Abduction has many useful applications in verification, including inference of missing preconditions, error diagnosis, and construction of compositional proofs. This paper gives a brief tutorial introduction to Explain and describes the basic inference algorithm.