CALAIS - A Component Analysis Learning Algorithm for Inner Source Development

Abstract

In the ever evolving world of software development, the complexity of products is increasing. This increased complexity is due to the integration of components built using multiple technologies. In this environment, companies are turning to open source software components to reduce software development time. These freely available open source components are often tried and tested by the software development community. Similar to open sourcing, inner sourcing involves the reuse of software components from other sections within large organizations. As with open sourcing, inner sourcing is experiencing a high adoption. Companies such as Philips, PayPal and Ericsson use open source software in an internal capacity to encourage the reuse of components. The challenge for system architects considering inner sourced components is to (a) determine the complexity, reliability, usage and therefore the importance of individual components within an overall product (b) assess the impact and importance of any individual component when components can differ in scale and technology. This work proposes CALAIS - A Component Analysis Learning Algorithm for Inner Source Development. CALAIS is a self-directed artificial neural network which uses historic performance to weigh the relative importance of an individual component within a system architecture. CALAIS operates by analyzing complexity, reliability, and usage. Using CALAIS, system architects can gain a fine grained view of the structural relevance of all system components proposed for inner sourcing. This view can be used to promote the delivery of high quality components within an inner source project.