Meta-Heuristic Search Based Test Data Generation Method for Dynamic-Functional Testing in Automatic Programming Assessment (S/O 13793)

• Main Authors: Romli, Rohaida, Abd Wahab, Alawiyah, Romli, Rusnida, Yusoff, Nooraini
• Format: Monograph
• Language: English
• Published: UUM
• Subjects: T Technology (General)
• Online Access: https://repo.uum.edu.my/id/eprint/31510/1/13793.pdf; https://repo.uum.edu.my/id/eprint/31510/
• Institution: Universiti Utara Malaysia

Automatic Programming Assessment (APA) has been widely known in the area of Computer Science education as a significant method in realizing automated marking and grading on students’ programming exercises. APA practically depends upon a test data generation process to perform a dynamic testing. Recently in software testing research, the adoption of any Meta-Heuristic Search Techniques (MHSTs) has successfully been proven to improve the efficiency of generating adequate and optimal test data. Unfortunately, current studies on APA have not yet usefully embrace the techniques to include better quality program testing coverage by considering the optimal size of the generated test data. Generating a large volume of test data has been identified as leading to issues concerning combinatorial problems. Thus, this research aims to propose a method of generating and locating an adequate test data with optimal in size by adapting a MHST to satisfy the dynamic functional testing in APA namely DyFunFPA-TDG. Initially, a comparative evaluation that mainly aims to compare among the most recent MHSTs so as to identify any applicable technique (s) to support automated test data generation in executing a dynamic-functional testing in APA was conducted. Based on the finding of this comparative study, the DyFunFPA-TDG method was constructed. A controlled experiment that adopts one-group pre-test and post-test design was carried out to evaluate DyFunFPA-TDG method in terms of the criteria of reliability and validity test data adequacy (or called positive and negative testing criteria). The findings reveal that DyFunFPA-TDG method is able to improve the measured criteria in the context of programming assessments. DyFunFPA-TDG method is expectantly to assist the lecturers who teach introductory programming courses to derive and generate test data and test cases to perform APA regardless of having a particular knowledge of test cases design in conducting a structural testing. By utilizing this method as part of APA, the lecturers’ workload can be reduced effectively since the typical manual assessments are always prone to errors and leading to inconsistency. This research contributes an improved test data generation method to assist educators of elementary programming courses to provide the means of deriving and generating adequate test data with optimal in size regardless of having the expertise in specific knowledge of test cases design