Test data adequacy measurement

Abstract

Criteria for test data adequacy are rules for deciding when enough testing has been performed. Development of these criteria is an important objective of research in software testing. In this paper, the abstract properties of adequacy criteria are investigated in a formal axiomatic style. Starting from work by Weyuker [1, 2] and Parrish and Zweben [3], we define an adequacy criterion as a function from specifications, programs and subsets of data to a degree of the adequacy of the test set. In this paper, we address the general properties of adequacy criteria, independent of the syntax and semantics of specifications and programs. In subsequent work, we will address properties related to specifications and programs. The dependencies between the axioms, as well as the relationship between them and those in Weyuker's and Parrish and Zweben's system, are also investigated.