Techniques for merging views of software processes

Abstract

Models of software development processes have many uses such as an aid to understanding, composing, assessing, improving and automating workflows. However, eliciting descriptive models from actual work environments can be quite complex due to multiplicity of roles, activities, artefacts, conditions, distributivity, locations and others. One way to manage this complexity is to elicit, from different sources, partial models (called views) of the subject process and then merge them into a coherent whole. In this paper, we describe "view-merging" algorithms, which form a core part of a view-based model elicitation system. The algorithms aid in identifying overlaps and inconsistencies and in presenting possible resolutions which, by interacting with the model elicitor, leads towards an incrementally built, unified, coherent process model. These algorithms have been implemented in a system called V elicit, which has been validated empirically. © 2010 Springer-Verlag Berlin Heidelberg.