The term Semantic Web denotes a vision of a new World-Wide Web in which different kinds of resources (data, services, Web pages, etc.) are accessed and shared on the basis of formal representation structures [7]. The main idea of the Semantic Web is to answer different kinds of queries w.r.t. a specific conceptual data model rich enough in expressivity to adequately describe the interrelationships between basic representational units (names, symbols) on which a piece of software for a certain application domain is based. Since symbols might have different meanings in different software applications, it is important to explicitly refer to meaning definitions given as a certain instance of an underlying conceptual data model. In the context of the Semantic Web, these definitions are seen as resources themselves and, for reasons of brevity, are often called ontologies. Ontologies provide the basis for implementing specific Semantic Web resources for encapsulating computational processes. These resources are called services [19]. In order to be unambiguous and suitable for machine processing, ontologies must be defined with reference to a formal semantics. A logical semantics provides the basis not only for the meaning of the representational language but also for the meaning of the associated query and command language. In many applications the logic behind conceptual data models must be decidable but very expressive [1] such that, in principle, query answering involves solving nontrivial logical inference problems. This article briefly introduces the foundation of Se mantic Web ontology representation languages, namely description logics [2]. For actually building systems for the Seman tic Web, however, we consider practical aspects of inference systems and their integration into a distributed systems context. Using the Racer System [16] as an example the paper sketches how description logic inference engines can be used to implement, for example, information retrieval services on the Semantic Web.
展开▼
机译：语义网一词表示对新型万维网的一种构想，在这种万维网中，根据形式表示结构，可以访问和共享各种资源（数据，服务，网页等）[7]。语义网的主要思想是用w.r.t.回答各种查询一个特定的概念数据模型，其表达能力足够丰富，足以充分描述基本表示单位（名称，符号）之间的相互关系，而基本表示单位（用于某个应用程序域的软件）所基于的关系。由于符号在不同的软件应用程序中可能具有不同的含义，因此重要的是明确引用作为基础概念数据模型的特定实例给出的含义定义。在语义网的上下文中，这些定义被视为资源本身，并且出于简短的原因，通常被称为本体。本体为实现用于封装计算过程的特定语义Web资源提供了基础。这些资源称为服务[19]。为了明确并适合于机器处理，必须参考形式语义来定义本体。逻辑语义不仅为表示语言的含义提供了基础，而且为关联的查询和命令语言的含义提供了基础。在许多应用中，概念数据模型背后的逻辑必须是可决定的，但必须具有很高的表达力[1]，因此，原则上，查询回答涉及解决非平凡的逻辑推理问题。本文简要介绍了语义Web本体表示语言的基础，即描述逻辑[2]。但是，对于实际构建用于语义Web的系统，我们考虑推理系统的实际方面以及将其集成到分布式系统上下文中。本文以Racer System [16]为例，描述了如何使用描述逻辑推理引擎来实现语义Web上的信息检索服务。
展开▼
