A Model-based Approach for Software Test Process Improvement

Abstract

Reducing the effort or cost of testing is a key issue to high productivity in software development. This paper proposes a test process model which aims to grasp the current status of test processes and to provide several metrics to evaluate efficiency of test processes in order to construct the process improvement plan. The model defines test process as an iterated process of four kinds of hypotheses development and verification which deal with: correctness of the program to be tested, location of failure, location of faults, and correction of the faults, respectively. The analysis results of experimental data collected from the test process show the validity of the proposed model and usefulness of the metrics.