Applying sampling to improve software inspections

Summary, in English

The main objective of software inspections is to find faults in software documents. The benefits of inspections are reported from researchers as well as software organizations. However, inspections are time consuming and the resources may not be sufficient to inspect all documents. Sampling of documents in inspections provides a systematic solution to select what to be inspected in the case resources are not sufficient to inspect everything. The method presented in this paper uses sampling, inspection and resource scheduling to increase the efficiency of an inspection session. A pre-inspection phase is used in order to determine which documents need most inspection time, i.e. which documents contain most faults. Then, the main inspection is focused on these documents. We describe the sampling method and provide empirical evidence, which indicates that the method is appropriate to use. A Monte Carlo simulation is used to evaluate the proposed method and a case study using industrial data is used to validate the simulation model. Furthermore, we discuss the results and important future research in the area of sampling of software inspections.

• Author: Thomas Thelin; Håkan Ahrefors; Per Runeson; Claes Wohlin
• Department/s: Department of Computer Science; Departments at LTH; Software Engineering Research Group
• Publishing year: 2004
• Language: English
• Pages: 257-269
• Publication/Series: Journal of Systems and Software
• Volume: 73
• Issue: 2
• Links: http://dx.doi.org/10.1016/S0164-1212(03)00249-8
• Document type: Journal article
• Publisher: Elsevier
• Topic: Computer Science
• Keywords: software inspection; Monte Carlo simulation; empirical study; efficiency; sampling
• Status: Published
• ISBN/ISSN/Other: ISSN: 0164-1212