Identification method in software process improvement areas

Abstract

With the prevalence of CMMI (Capability Maturity Model Integration) that is developed in the United States and the international standardization model of Software Process Assessment ISO/IEC 15504, SPI (Software Process Improvement) based on Software Process Assessment is gaining ground in Japan as well. One of the objectives in SPI is to prevent the occurrence of errors in the downstream process by thorough process management in the upstream process. In order to implement the SPI from such viewpoints, this paper suggests the method to analyze Problem Reports developed in the testing process of a project, to specify software bugs existent in leading processes and to identify the SPI in the upstream process. By means of Problem Reports that is capable of externalization and is used in testing processes, knowledge that has not been externalized will be specified as Defect Cluster. Furthermore, deliverables in each process will be identified from the properties of Defect Cluster to implement the SPI, which is of the essence, by identifying processes causing trouble. In the project that was the analytical target, there were 608 Problem Reports in its testing process. As a result of the analysis, it was found that by identifying processes in the upper process to improve with the techniques recommended in this paper, about 13% of cases causing trouble could be prevented from occurring, which will contribute to productivity improvement due to a decrease in the number of rework processes. © 2008 The Institute of Electrical Engineers of Japan.