Effective Regression Testing Using Requirements and Risks

Abstract

The use of system requirements and their risks enables software testers to identify more important test cases that can reveal faults associated with risky components. Having identified those test cases, software testers can manage the testing schedule more effectively by running such test cases earlier so that they can fix faults sooner. Some work in this area has been done, but the previous approaches and studies have some limitations, such as an improper use of requirements risks in prioritization and an inadequate evaluation method. To address the limitations, we implemented a new requirements risk-based prioritization technique and evaluated it considering whether the proposed approach can detect faults earlier overall. It can also detect faults associated with risky components earlier. Our results indicate that the proposed approach is effective for detecting faults early and even better for finding faults associated with risky components of the system earlier than the existing techniques.

Risk-based test case prioritization using a fuzzy expert system

“…In order to perform risk assessment for the requirements, we identify four risk indicators that have been used by previous requirements and risk-based regression testing research [6,21,25,41]: requirements complexity (RC), requirements size (RS), requirements modification status (RMS), and potential security threats (PST). These previous studies indicate that these risk indicators can be effective in finding defects in software systems, thus we focus on these four risk indicators in this study, but we consider to use other risk indicators, such as usage rate [16], as we evaluate our approach in the future.…”

“…We improved this process by introducing weights for the risk items (RiIM). The weight values for the risk items were obtained from our previous requirements riskbased research [21], and for this research, we further calibrated the weight values by using the risk exposure values for risk items in Yoon [48]. The risk items denoted potential defect types for a software system.…”

“…reordered test cases detected faults; the approaches utilized risk information to prioritize tests, so they should be evaluated by measuring whether the detected faults are, indeed, from the locations where risks reside in the product. To address these limitations, our previous work [21] proposed a new requirements risk-based test case prioritization approach by considering the direct relationship between requirements risks and the test case. We also introduced a new evaluation method to measure how effective test case prioritization approaches were for detecting defects in risky components of software systems.…”

Coverage-Based Clustering and Scheduling Approach for Test Case Prioritization

“…Yoon et al [24] identify more important test cases that are likely to detected effects associated with the system's risks by using requirements risks. Hettiarachchi et al [25], [26] propose a new requirements risk-based test case prioritization approach by considering the direct relationship between requirements risks and the test case.…”

Recent Advances in Regression Testing Techniques

“…Arafeen and Do [57] proposed an approach that clusters requirements based on similarities obtained through a text-mining technique and that prioritizes test cases using the requirements-tests relationship. In addition to requirements and design information, some other researchers have used software risk information to prioritize test cases in order to exercise test cases on the code areas with potential risks as early as possible [58,59].…”