Generating an Automated Test Suite by Variable Strength Combinatorial Testing for Web Services

Journal of Computing and Information Technology. 2016;24(3):271-282 DOI 10.20532/cit.2016.1002727

 

Journal Homepage

Journal Title: Journal of Computing and Information Technology

ISSN: 1846-3908 (Online)

Publisher: University of Zagreb Faculty of Electrical Engineering and Computing

Society/Institution: University of Zagreb Faculty of Electrical Engineering and Computing

LCC Subject Category: Science: Mathematics: Instruments and machines: Electronic computers. Computer science

Country of publisher: Croatia

Language of fulltext: English

Full-text formats available: PDF

 

AUTHORS

Yin Li (Jiangsu Institute of Automation, Jiangsu, Lianyungang, Xin Pu Qu, China)
Zhi-an Sun (Jiangsu Institute of Automation, Jiangsu, Lianyungang, Xin Pu Qu, China)
Jian-Yong Fang (Jiangsu Institute of Automation, Jiangsu, Lianyungang, Xin Pu Qu, China)

EDITORIAL INFORMATION

Blind peer review

Editorial Board

Instructions for authors

Time From Submission to Publication: 26 weeks

 

Abstract | Full Text

Testing Web Services has become the spotlight of software engineering as an important means to assure the quality of Web application. Due to lacking of graphic interface and source code, Web services need an automated testing method, which is an important part in efficiently designing and generating test suite. However, the existing testing methods may lead to the redundancy of test suite and the decrease of fault-detecting ability since it cannot handle scenarios where the strengths of the different interactions are not uniform. With the purpose of solving this problem, firstly the formal tree model based on WSDL is constructed and the actual interaction relationship of each node is made sufficient consideration into, then the combinatorial testing is proposed to generate variable strength combinatorial test suite based on One-test-at-a-time strategy. At last test cases are minimized according to constraint rules. The results show that compared with conventional random testing, the proposed approach can detect more errors with the same amount of test cases which turning out to be more ideal than existing ones in size.