The Metrics Apprentice: using cultural algorithms to formulate quality metrics for software systems

The Metrics Apprentice processes a domain knowledge base of software quality concepts with a form of evolutionary computation in order to learn software metrics for a delimited application or software development environment. The evolutionary computation method that is used, the Cultural Algorithm, uses beliefs about the performance of individual population members in order to enhance the evolutionary learning process. In the Metrics Apprentice, these beliefs are an integrated part of the domain knowledge base, and the ones that are most useful in the learning process persist for reuse in future learning tasks. The semantic network that encodes the domain of software quality issues and concepts is displayed using an extension of expandable outlines called the Outline Knowledge Display.