Abstract program slicing on dependence condition graphs

Context: Mastroeni and Zanardini introduced the notion of semantics-based data dependences, both at concrete and abstract domains, that helps in converting the traditional syntactic Program Dependence Graphs (PDGs) into more refined semantics-based (abstract) PDGs by disregarding some false dependences from them. As a result, the slicing techniques based on these semantics-based (abstract) PDGs result in more precise slices. Aim: The aim of this paper is to further refine the slicing algorithms when focusing on a given property. Method: The improvement is obtained by (i) applying the notions of semantic relevancy of statements and semantic data dependences, and (ii) combining them with conditional dependences. Result: We provide an abstract slicing algorithm based on semantics-based abstract Dependence Condition Graphs (DCGs) that enable us to identify the conditions for dependences between program points.