PAntUDE - an anti-unification algorithm for expressing refined generalizations

Abstract

In this paper some improvements for the basic algorithm for anti-unification are presented. The standard (basic) algorithm for anti-unification still may give too general answers with respect to the intended use of the result. "Too general" means obtaining an unwanted answer when instantiating some variables of the anti-unification output term. To avoid this, a term sometimes should be generalized by regarding certain (semantic) restrictions. In PAntUDE (partial anti-unification with domains and exclusions) two principal improvements of the basic algorithm were implemented and tested: it can use masks for preventing anti-unification of certain arguments (partial anti-unification); it can also use finite domains for enumerating input-term constants instead of introducing a new (universally quantified) variable, and finite exclusions for specifying forbidden constants (anti-unification with domains and exclusions).