An Industrial Study of Natural Language Processing Based Test Case Prioritization

Abstract

In mobile application development, the frequentsoftware release limits the testing time resource. In order todetect bugs in early phases, researchers proposed various testcase prioritization (TCP) techniques in past decades. In practice, considering that some test case is described or contains text, theresearchers also employed Natural Language Processing (NLP)to assist the TCP techniques. This paper conducted an extensiveempirical study to analyze the performance of three NLP basedTCP technologies, which is based on 15059 test cases from 30industrial projects. The result shows that all of these threestrategies can help to improve the efficiency of software testing, and the Risk strategy achieved the best performance across thesubject programs.