Bending the Rules: Steps Toward Semantically Enriched Graphical Rewrite Rules

dimensional situations containing objects (Figure 1). Situations can be interpreted with respect to objects contained and spatial relationships holding between these objects. The differences between situations imply one or more actions capable of transforming one situation into another. Figure 1 depicts a rewrite rule containing cars, traffic lights, and roads. One possible interpretation for the action implied is that the car moved. Graphical rewrite rules, as a form of end-user programming, suffer from their implicit underlying model. Interpretation of rewrite rules limited to syntactic properties makes it laborious for end users to define non-trivial behavior. Semantically enriched graphical rewrite rules have increased expressiveness, resulting in a significantly reduced number of rewrite rules. This reduction is essential in order to keep rewrite rule-based programming approaches feasible for end-user programming. The extension of the rewrite rule model with semantics not only benefits the definition of behavior but additionally it supports the entire visual programming process. Specifically the benefits include support for defining object look, laying out scenes consisting of dependent objects, defining behavior with a reduced number of rewrite rules, and reusing existing behaviors via rewrite rule analogies. These benefits are described in the context of the Agentsheets programming substrate. Figure 1. Rewrite Rule = ⇒ Action(s) The implicit nature of graphical rewrite rules, due to the lack of sufficient interpretational clues, makes it hard for end users to assess their general applicability. Would the rule shown also work for other cars, other roads? To what other related situations would the rule apply? Is the car in situation 1 the same car as in situation 2? Unless rewrite rules serve as literal substitutions of one situation with another it is important for the end user to understand how rules are interpreted in order to allow for generalization of objects and spatial relationships between objects.