Protocol Models of Human-Computer Interaction

Abstract

Abstract:Current approaches to modeling human-computer interaction do not always succeed in producing behaviorally complete models of manageable size and complexity. We argue that the reason for this lies in lack of support for parallel composition of partial behavioral descriptions, and propose the recently developed Protocol Modeling approach as a superior alternative. The semantics of Protocol Modeling support separation and composition of concerns in models of human-computer interaction, and the production of executable models to explore and refine the desired behavior. Protocol Modeling supports a crucial property sought in modeling methods if it is to scale to complex problems, namely the ability to reason about the modeled behavior of the whole based on examination only of a part (sometimes called "modular" or "local" reasoning).