Analysis of fault/failure data from student software development projects.

摘要

Due to the increasing complexity of software systems, it has become extremely difficult to ensure software quality in a cost-effective manner. The National Institute of Standards and Technology (NIST) project entitled "Reference Data: Software Error, Fault, Failure Data Collection & Analysis Repository Project" hopes to address this problem. The project serves to benefit industry by creating a public repository for software fault data. Contributed data will be provided through various industrial, government, and academic sources. This thesis focuses on the efforts of one such academic partner, the University of Louisville (UL). UL targeted three upper-level development courses as the source for data to be contributed to the EFF project. By validating and summarizing the collected data, checking it for consistency, and offering suggestions regarding tool improvements and study extensions, the University of Louisville intends to assist NIST in meeting its goals for the EFF project.