链式桥规则导致的D3L知识传播及推理

Knowledge Propagation and Reasoning Induced by Bridge Rule Chains in D3L

作为分布式、模块化本体的逻辑基础,分布式动态描述逻辑(D3L)使得通过有向语义映射连接的多个动态描述逻辑本体之上的推理成为可能.文中针对复合桥规则的知识传播进行了研究.作者发现当桥规则构成合取链的情况下,D3L的原始语义不能保证本体元素的局部语义关系正确传播到目标本体.通过引入合取映射桥规则,作者对D3L进行了扩展.由于对合取映射对应的目标元素属性进行了约束,新的桥规则可以支持局部语义关系在分布式本体间的正确传播,并且不改变D3L的原始属性.最后提出了合取映射桥规则到传统桥规则的转换语义,并证明了基于该语义的Tableaux推理算法的正确性.扩展后的D3L更好地实现了在本体间导入和重用知识的思想,因此可以为语义Web提供较完备的逻辑基础.

Abstract As a logical foundation for distributed and modular ontology, Distributed Dynamic Description Logics （D3L） enables reasoning with multiple ontologies interconnected by directional semantic mapping （bridge rules）. In this paper, we study knowledge propagation in the case of complex bridge rules exist. We find the local semantics of ontology elements cannot always propa- gates to the target ontologies if bridge rules form conjunctive chain. By inducing conjunctive mapping bridge rules, we modify the original semantics of D3L in order to cope with this modeling discrepancy. If conjunctive bridge rules are using instead of the normal ones, the problem does not occur any more. New semantics enable sound and complete propagation of the local semantics between distributed ontologies. All the basic properties that have been established for D3L hold also for the adjusted framework. We also provide a transformational semantics for conjunctive mapping bridge rules and give a proof for the correctness of Tableaux reasoning algorithm based on the new semantics. The D3L with conjunctive mapping bridge rules can capture the idea of importing and reusing knowledge between ontologies better and combine well with intuitions behind the Semantic Web, thus provide a complete logical basis for the Semantic Web.