Classifying software process models based on natural language descriptions

Reuse of the valuable knowledge gained through the realization of software projects is an important step in overcoming well-known problems of the software industry such as wrong schedules and cost estimations, low productivity, and low product quality. To promote this kind of reuse the authors propose an approach relying on an explicit representation of the software processes by process descriptions and to organize and classify them in a software process library for further use within software projects. A process description is divided into an informal process definition document and a formal process model. The process definition document constitutes the basis for the classification process that is performed by using an artificial neural network. They present an exposition of their approach and discuss the promising results of a case study in structuring a software process library.