A Roadmap for Rules and RuleML in the Semantic Web

RuleML is the leading approach to Semantic Web rules: Over the last two and a half years or so, a broad consensus has evolved in the Semantic Web community that the vision of the Semantic Web includes, specifically, rules – as well as ontologies cf. OWL. RuleML has emerged as the leading current standardization approach to rules for the Semantic Web. We are a co-founder and co-chair of the RuleML Initiative, and co-chair of the closely related DAML Rules effort which is currently the main focus of the Joint Committee that developed DAML+OIL. In this paper, we give a roadmap for rules in the Semantic Web, including overview, design rationale, and current directions of RuleML. Declarative knowledge representation: Standards should be founded upon techniques that are well established at a research level. The only approach to making the Semantic Web be " Semantic " that is well understood – and well accepted – from a substantial body of previous research is that it must be founded upon declarative knowledge representation (KR). RuleML, like OWL and RDF, shares this approach. To declarative KR, each then adds Webized syntax. ``Declarative '' here means in the sense of (AI) KR theory. A given set of premises entails a set of sanctioned conclusions, independent of inferencing control strategy or procedural aspects, e.g., independent of whether inferencing direction is goal-directed query answering (``backward'') vs. data-driven (``forward''). Declarativeness greatly facilitates reuse, and multiple kinds of uses, of knowledge. It enables one knowledge-based application (i.e., " agent ") to anticipate precisely and completely what meaning will be drawn by anotheragent from knowledge that has been communicated (e.g., shared). " Webized " here means using the Web's overall open spirit and standards suite – in particular XML, URI's, and namespaces – so as to support modules of multiply authored, widely distributed knowledge. RuleML has initially emphasized XML as the form of syntax, to facilitate building of tools for translation and inferencing. But it also has drafted an RDF syntax and a human-oriented string syntax (as well as an abstract syntax that bridges all these). However, for rules, as for ontologies, the most crucial design choice is the choice of underlying fundamental knowledge representation. Declarative logic programs as KR shared by commercially important rule systems: There are four families of rule systems that are the most currently commercially important (``CCI''): SQL (relational database), Prolog, production rules (cf. OPS5, CLIPS, …