An Empirical Analysis of Modal Theorem Provers

Hustadt, U. and Schmidt, R. A. (1999)

Journal of Applied Non-Classical Logics 9(4) 479-522. BiBTeX, PostScript.

This paper reports on an empirical performance analysis of four modal theorem provers on benchmark suites of randomly generated formulae. The theorem provers tested are the Davis-Putnam-based procedure KSAT, the tableaux-based system KRIS, the sequent-based Logics Workbench, and a translation approach combined with the first-order theorem prover SPASS.
Our benchmark suites are sets of multi-modal formulae in a certain normal form randomly generated according to the scheme of Giunchiglia and Sebastiani [GS96a, GS96b]. We investigate the quality of the random modal formulae and show that the scheme has some shortcomings, which may lead to mistaken conclusions. We propose improvements to the evaluation method and show that the translation approach provides a viable alternative to the other approaches.

See also On Evaluating Decision Procedures for Modal Logics, in Proc. IJCAI'97.
Renate A. Schmidt
Home | Publications | FM Group | School | Man Univ

Last modified: 31 Oct 03
Copyright © 1996,7,8 Renate A. Schmidt, School of Computer Science, Man Univ, schmidt@cs.man.ac.uk