On knowledge management in software development life cycles

Abstract

Software engineering is a knowledge-intensive activity. For software organisations, the main assets are not manufactur- ing plants, buildings, and machines, but the knowledge held by the employees. Studies [13][19] have shown that projects do not tend to fail because of developers' lack of technical knowledge, but rather for reasons such as requirements fail- ures, communication failures and estimation failures. These failures can be traced back to inadequate knowledge man- agement practices as a root cause. Software development processes tend to address knowledge management issues by prescribing documentation, a strategy which to knowledge management practitioners is known as codi cation. Codi - cation is the process of converting a body of knowledge held within a `knower' (tacit knowledge) to a form which makes the knowledge permanent and thus easier to transfer (ex- plicit knowledge). As with most strategies/techniques, this approach works in some cases but less so in others. As software systems become increasingly complex and dead- lines increasingly tight, this paper argues that the profession will bene t if software development life cycles evolved to explicitly handle knowledge management aspects. The ap- proach taken here is to provide an overview of some core knowledge management concepts whilst giving examples of how they apply (or not) to the software engineering world. The material presented here is related to ongoing Ph.D. re- search within the department entitled \A knowledge driven software development life cycle".