Test Selection and Coverage Based on CTM and Metric Spaces

One of the main open issues in testing and in particular in conformance testing is finding a good way of measuring the quality of the testing activity. With other words after performing a given quantity of tests, how much it is covered from the number of potential errors of the implementation under test (IUT). The present paper tries to build up a solution for this necessity. In (L. Feijs et al., 2002, N. Goga et al., 2004) we proposed a general framework for defining a coverage measure based on a distance measure between different behaviors of a system under test. In the current paper we apply this general theory to another domain, namely to CTM (classification tree method) that is usually used for the selection of the data for testing. As a case study we apply this technology for testing a software tool used for computer music generation