Automatically generating test inputs for components\udwithout source code (are ‘black-box’) and specification is challenging.\udOne particularly interesting solution to this problem is to\uduse Machine Learning algorithms to infer testable models from\udprogram executions in an iterative cycle. Although the idea has\udbeen around for over 30 years, there is little empirical information\udto inform the choice of suitable learning algorithms, or to show\udhow good the resulting test sets are. This paper presents an\udopenly available framework to facilitate experimentation in this\udarea, and provides a proof-of-concept inference-driven testing\udframework, along with evidence of the efficacy of its test sets on\udthree programs.
展开▼
