Detecting Duplicate Actions in a KDT Script Based on LRS and LCS

劉建宏(CHIEN-HUNG LIU) ； 陳偉凱(WOEI-KAE CHEN) ； 廖振諺(CHEN-YAN LIAO)

《Journal of Information Science and Engineering》 36卷3期 (2020/05) Pp. 513-533

https://doi.org/10.6688/JISE.202005_36(3).0003

摘要

In keyword-driven testing (KDT), having duplicate actions in the test script is perhaps the most common bad smell. Once the target user interface is changed, a KDT script with duplicate actions can be difficult to maintain. Thus, detecting and removing duplicate actions is an important task. However, so far, no KDT testing tools support automated duplicate detection. This paper proposes a method and tool, called DDT (Duplicate script Detection Tool), for the tester to quickly identify duplicate actions. Two detection algorithms based on Longest Repeated Substring (LRS) and Longest Common Subsequence (LCS) are presented. In addition, DDT provides a keyword extraction feature that can automatically remove duplicate actions. Our experimental results show that there are 21-42% of duplicate actions in a typical KDT script, DDT can detect these duplicate actions in 3-5 seconds, and up to 58-81% of these duplicate actions should be refactored.

關鍵字

duplicate detection ； keyword extraction ； bad smell ； KDT scripts ； duplicate actions