Inferring Automatic Test Oracles
William B. Langdon, Shin Yoo, Mark Harman · 2017
We propose the use of search based learning from existing open source test suitesto automatically generate partially correct test oracles. We argue that mutation testing and n-version computing(augmented by deep learningand other soft computingtechniques), will be able to predict whether a program's output is correct sufficiently accurately to be useful.