Elements of a First Visual Rule Language for the Semantic Web

Summary Our survey in the previous sections shows that there are many possibilities to visualize rules in different ways with different purposes. We have argued that a graphical notation for rules in UML class diagrams would be of particular importance, since class diagrams visualize vocabularies (resp. ontologies) and rules are based on vocabularies. Such a notation would therefore allow showing: • how rules are based on vocabularies • how rules extend ontologies • how rules can be used in a model-driven software engineering approach such as OMG’s Model-Driven Architecture (MDA) Based on our survey we can identify several requirements for such a visual rule language. In particular, we need: -a graphical symbol for rules: Rule is a new metaclass to be added to the UML metamodel. A graphical symbol for rules should have a shape that is different from all other model element shapes currently used in UML class diagrams. -a (semi-)graphical notation for certain types of logical formulas expressed in terms of other class modeling elements. -a notation for event types. For visualizing reaction rules, we need to distinguish two kinds of events: • action events, which can be the result of firing a reaction rule that can trigger a reaction rule • non-action events, such as time events, which cannot be the result of firing a reaction rule but which can trigger a reaction rule Logical formulas may play the role of conditions, conclusions and postconditions in a rule. In each role, their syntax may be specifically restricted. It seems to be natural to use the UML symbol for states, a rectangle with rounded corners for expressing: • status predicate conditions and conclusions by using the name of the Boolean attribute as the name of the state rectangle • general state (post-)conditions by putting a Boolean OCL expression in the state rectangle (instead of a name) How to express other types of atomic formulas is an issue for further research. This preliminary report is necessarily incomplete and leaves many issues unsolved. In particular, it would be important: • to relate rule modeling and visualization to the three abstraction levels defined by the Model-Driven Architecture for the Object Management Group • to integrate our rule modeling concepts with the UML metamodel • to investigate what are the specific issues of modeling and visualizing Semantic Web rules based on RDF and OWL