Difference Measure for Controlled Random Tests

Abstract

The task of constructing test sequences difference characteristics was studied. Its relevance for generating controlled random tests and complexity in finding difference measures for the case of symbolic tests were substantiated. The limitations of using traditional distance characteristics to obtain a measure of the difference between test sets are shown. For the binary case, a new measure of the difference MH(Ti, Tk) of two character test sets Ti and Tk is defined based on the classical Hamming distance. This measure represents n components, each of which is determined by the Hamming distance between the binary set Ti and the pattern Tk cyclically shifted by v bits. The main properties of the proposed dissimilarity measure are reviewed and its effectiveness for classifying test candidates when generating controlled random tests is shown. Experimental results are presented that confirm the effectiveness of the proposed difference measure.

Keywords

• measure of difference
• hamming distance
• levenshtein distance
• test
• test pattern