Expressing and Managing Reactivity in the Semantic Web

Ontological knowledge and reasoning provide the basis to define the semantics and static properties of Web resources. However, in existing approaches data reactivity is encoded in a different formalism by using rules, and the integration of ontologies and rules is not always natural or user-friendly. In this paper we present an alternative approach to represent active knowledge in ontologies. First, the ACTION formalism, in which events are categorized as ontological concepts, is proposed. Events are used in conjunction with classes, properties and instances during reasoning tasks and query answering. ACTION ontologies are processed within the REACTIVE framework. The REACTIVE reasoning and query engine supports the discovery tasks required to identify the effects of a given set of fired events. Additionally, an optimization strategy named IMR (Intersection of Magic Rewritings) is implemented. IMR identifies the events and properties that need to be considered multiple times and constructs the minimal set of rules that will produce the required result. The expressiveness of the ACTION formalism was empirically studied as well as the performance of the optimization and evaluation strategies. Initial experimental results suggest that ACTION is more expressive than rule-based formalisms; in addition, the REACTIVE engine in conjunction with IMR strategies reduce execution time to at least 50% of the execution time of traditional strategies.