From visual language specification to legal visual interaction

A crucial requirement of visual interaction is to prevent users from performing incorrect operations. This requires the computer environment to trap illegal users´ actions. On the other hand, users should have enough freedom in their interaction, typically by not being forced to follow predefined sequences of operations, yet being able to develop their line of thought in solving their problems. The paper argues that an interactive session can be formalised as a set of legal visual sentences, i.e., a visual language specified via visual conditional attributed rewriting systems (vCARWs), and that this formalisation allows the specification of the control part of a visual interactive system which traps illegal users´ actions, still leaving them freedom in the order of actions. We discuss two algorithms to derive an automaton controlling the interaction from the specification of a vCARW