Towards a Novel Statistical Method for Generating Test Sets with a Given Coverage Probability

This paper presents a method that uses a small set of execution samples to select a minimal set of execution paths, which has the property of its coverage probability being above a required confidence level, and then generate a natural language specification of the test case set. Experimental results show that it is not only simple to be applied but also generates a reliable test case set. The greatest benefit, however, is the use of a minimal set of test cases, which reduces the computational effort to guarantee that the program reliability is above a required level of confidence.