Effects of Error Detection and Representation on the Resolution of Traceability Issues

Abstract

Traceability is extremely important when developing safety critical software systems. The automotive industry has strict regulations that require developers to be able to demonstrate traceability links from requirements to end products. Keeping traceability links up to date is a very expensive task; both financially and in terms of man hours. This paper conducts research into the effects of error detection and representation when traceability links become inconsistent. Additionally, we look at the effects of automating some of the process for fixing inconsistencies. The research is conducted by performing 2 experiments using a traceability management tool developed at Gothenburg University. The first investigates the use of notifications of issues and the second the use of automation when fixing issues. Our results showed that with notifications in small projects there is no significant difference when compared to finding and fixing issues manually. Our second experiment showed a significant difference in time when using automation to assist with fixing issues compared to correcting issues manually.