Inspection effectiveness in software development: a neural network approach

In this paper we discuss an analysis of inspection effectiveness based on defect escapes. We present a neural network approach to inspection based on the back propagation model for identifying inspections with defect escapes. Our analysis shows several findings that provide new insights on defect escapes and inspection effectiveness. Our approach is quite novel, not only because of its focus on defect escapes, but also because of its application of neural network techniques to the analysis of software inspection effectiveness. We believe that other software development organizations may benefit from this work as well.