Answer Set Programming with Composed Predicate Names

Mainstream systems for Answer Set Programming implement intelligent grounding to eliminate object variables from the input program, often obtaining a propositional program of reasonable size. However, non-stratified negation may inhibit the simplification of some rule bodies due to the lack of knowledge on the truth of recursive atoms. Frustration is greatest when the program is clearly locally stratified, such as in case of numerical arguments in rule heads obtained by increasing some body arguments; common examples are minimal distances in graphs and time arguments in planning scenarios. This paper suggests to move some arguments in predicate names, so that the declarative semantics of Answer Set Programming is preserved, but non-stratified negation is possibly avoided thanks to symbolic rule instantiation. A proof of concept is given in terms of Jinja templates for arguments with a clear range.