On Syntactic Forgetting with Strong Persistence

Abstract

It is generally agreed upon that so-called strong persistence (SP) captures best the essence of forgetting in logic programming. While classes of operators, such as F_R and F_SP, that satisfy immediate relaxations of (SP), and in the case of F_SP, even (SP) whenever possible, have been characterized semantically, many practical questions have yet to be addressed. This technical paper aims to answer one of them: How can atoms be forgotten from a program without having to calculate its exponential number of models? To this end, we introduce two concrete representatives of F_R and F_SP that forget sets of atoms by syntactical manipulation of a program’s rules. This may in many cases prevent exponential blowups and produce a forgetting result that is close to the original program.