Design patterns for models of interactive systems

Abstract

Building models of safety-critical interactive systems (in healthcare, transport, avionics and finance, to name but a few) as part of the design process is essential. It is also advised for non-safety critical interactive systems if we want to be certain they will behave as intended in all circumstances. However, modelling interactive systems is also challenging. The levels of complexity in modern user interfaces and the wealth of interaction possibilities means that modelling at a suitable level of abstraction is crucial to ensure our models remain reasonably sized, readable, and therefore usable. The decisions we make about how to abstract the system to retain enough detail to be able to reason about it without running into known modelling problems (state-explosion, verbosity, unread ability) are complex, even for experienced modellers. We have identified a number of commonly seen problems in such models based on occurrences of common properties of interactive systems, and in order to help both experienced and novice modellers we propose model-patterns as a solution to this.