 
Summary: Maximizing Theory Accuracy
Through Selective Reinterpretation
Shlomo ArgamonEngelson argamon@mail.jct.ac.il
Department of Computer Science, Jerusalem College of Technology, Machon Lev,
P.O.B. 16031, 91160 Jerusalem, Israel
Moshe Koppel koppel@cs.biu.ac.il
Hillel Walters
Department of Mathematics and Computer Science, BarIlan University, 52900 Ramat
Gan, Israel
Abstract. Existing methods for exploiting awed domain theories depend on the use of a
su ciently large set of training examples for diagnosing and repairing aws in the theory.
In this paper, we o er a method of theory reinterpretation that makes only marginal use
of training examples. The idea is as follows: Often a small number of aws in a theory
can completely destroy the theory's classi cation accuracy. Yet it is clear that valuable
information is available even from such awed theories. For example, an instance with
several independent proofs in a slightly awed theory is certainly more likely to be correctly
classi ed as positive than an instance with only a single proof.
This idea can be generalized to a numerical notion of \degree of provedness" which
measures the robustness of proofs or refutations for a given instance. This \degree of
provedness" can be easily computed using a \soft" interpretation of the theory. Given a
