Abstract
Software inspection play a crucial role in achieving high quality software right from the beginning. Especially for requirements documents inspections are beneficial as defects can be detected and removed at an early point in time before they can leak into subsequent phases of the development process, where those defects can cause high rework cost and quality problems [Boehm and Basili, 2001].
In order to fully exploit the beneficial aspect of inspections it is necessary to control them: Based on an analysis of the inspection effectiveness (e.g., what type or number of defects have been detected and what type and number might be remaining) it has to be decided whether development can proceed as planned or controlling actions, such as reinspections, have to be performed.
Defect classification is a technique that can support this activity. Using appropriate defect classification schemes it is possible to get a quantitative statement of the inspection effectiveness (i.e., what type of defects have been found). Examples in the literature demonstrated beneficial applications but are hard to adopt without tailoring the underlying defect classification scheme to the environment.
In this report a defect classification scheme is developed that is to support the control of the inspection of QUASAR requirements documents. In particular we discuss our approach to define the scheme, its underlying rationale, the scheme itself, and our experience from using it in a case study.
The scheme consist of two defect attributes that are to be used when inspecting use cases and recording defects in these documents, and three attributes for state chart documents.

[]