A Reduction Result for Circumscribed Semi-Horn Formulas

  • Authors:
  • Patrick Doherty;Witold Łukaszewicz;Andrzej Szałas

  • Affiliations:
  • (Supported in part by the Swedish Council for Engineering Sciences (TFR)) Department of Computer and Information Science, Linköping University, S-581 83 Linköping, Sweden, e-mail: patdo@ ...;(Supported in part by KBN grant 3 P406 019 06) Institute of Informatics, Warsaw University, ul. Banacha 2, 02-097 Warsaw, Poland, e-mail: witlu@mimuw.edu.pl;(Supported in part by KBN grant 3 P406 019 06) Institute of Informatics, Warsaw University, ul. Banacha 2, 02-097 Warsaw, Poland, e-mail: szalas@mimuw.edu.pl

  • Venue:
  • Fundamenta Informaticae
  • Year:
  • 1996

Quantified Score

Hi-index 0.00

Visualization

Abstract

Circumscription has been perceived as an elegant mathematical technique for modeling nonmonotonic and commonsense reasoning, but difficult to apply in practice due to the use of second-order formulas. One proposal for dealing with the computational problems is to identify classes of first-order formulas whose circumscription can be shown to be equivalent to a first-order formula. In previous work, we presented an algorithm which reduces certain classes of second-order circumscription axioms to logically equivalent first-order formulas. The basis for the algorithm is an elimination lemma due to Ackermann. In this paper, we capitalize on the use of a generalization of Ackermann's Lemma in order to deal with a subclass of universal formulas called semi-Horn formulas. Our results subsume previous results by Kolaitis and Papadimitriou regarding a characterization of circumscribed definite logic programs which are first-order expressible. The method for distinguishing which formulas are reducible is based on a boundedness criterion. The approach we use is to first reduce a circumscribed semi-Horn formula to a fixpoint formula which is reducible if the formula is bounded, otherwise not. In addition to a number of other extensions, we also present a fixpoint calculus which is shown to be sound and complete for bounded fixpoint formulas.