Regression testing framework for test cases generation and prioritization

Abstract

A regression test is a significant part of software testing. It is used to find the maximum number of faults in software applications. Test Case Prioritization (TCP) is an approach to prioritize and schedule test cases. It is used to detect faults in the earlier stage of testing environment. Code coverage is one of the features of a Regression Test (RT) that detects more number of faults from a software application. However, code coverage and fault detection are reducing the performance of existing test case prioritization by consuming a lot of time for scanning an entire code. The process of generating test cases plays an important role in the prioritization of test cases. The existing automated generation and prioritization techniques produces insufficient test cases that cause less fault detection rate or consumes more computation time to detect more faults. Unified Modelling Language (UML) based test case generation techniques can extract test cases from UML diagrams by covering maximum part of a module of an application. Therefore, a UML based test case generation can support a test case prioritization technique to find a greater number of faults with shorter execution time. A multi-objective optimization technique able to handle multiple objectives that supports RT to generate more number of test cases as well as increase fault detection rate and produce a better result. The aim of this research is to develop a framework to detect maximum number of faults with less execution time for improving the RT. The performance of the RT can be improved by an efficient test case generation and prioritization method based on a multi-objective optimization technique by handling both test cases and rate of fault detection. This framework consists of two important models: Test Case Generation (TCG) and TCP. The TCG model requires an UML use case diagram to extract test cases. A meta heuristic approach is employed that uses tokens for generating test cases. And, TCP receives the extracted test cases with faults as input to produce the prioritized set of test cases. The proposed research has modified the existing Hill Climbing based TCP by altering its test case swapping feature and detect faults in a reasonable execution time. The proposed framework intends to improve the performance of regression testing by generating and prioritizing test cases in order to find a greater number of faults in an application. Two case studies are conducted in the research in order to gather Test Case (TC) and faults for multiple modules. The proposed framework yielded a 92.2% of Average Percentage Fault Detection with less amount of testing time comparing to the other artificial intelligence-based TCP. The findings were proved that the proposed framework produced a sufficient amount of TC and found the maximum number of faults in less amount of time.