The problem of managing the consistency of heterogeneous, distributed software engineering documents is central to the development of large and complex systems. We show how this problem can be addressed using xlinkit, a lightweight framework for consistency checking that leverages standard Internet technologies. xlinkit provides flexibility, strong diagnostics, and support for distribution and document heterogeneity. We use xlinkit in a comprehensive case study that demonstrates how design, implementation and deployment information of an Enterprise JavaBeans system can be checked for consistency, and rechecked incrementally when changes are made.

Software development often suffers from well-known problems such as wrong schedules and cost estimations, low productivity, and low product quality. In order to overcome these problems we suggest adapting the concepts of "organizational memory" and "organizational learning" and we argue in favor of establishing a reuse culture of software process models. We introduce an approach based on the process definition/instantiation/enaction paradigm and on the reuse of explicit software process descriptions (process models). The key features of our approach are the division of process descriptions into a goal-oriented process definition document and a formal implementation-oriented process model on the one hand and the use of an artificial neural network, more precisely a self-organizing map, for classification and retrieval purposes on the other. We present an exposition of our approach and discuss the promising results of an experiment in structuring a software process library and retrieving reuse candidates for upcoming projects.