The Semantic Web is an extension of the current World Wide Web, based on the idea of exchanging information with explicit, formal, and machine-accessible descriptions of meaning. Providing information with such semantics will enable the construction of applications that have an increased awareness of what is contained in the information they process and that can therefore operate more accurately. This has the potential of improving the way we deal with information in the broadest sense possible, for example, better search engines, mobile agents for various tasks, or even applications yet unheard of. Rather than being merely a vision, the Semantic Web has significant backing from various institutes such as DARPA, the European Union, and the W3C, all of which have performed a variety of Semantic Web activities. In order to be able to exchange the semantics of information, one first needs to agree on how to explicitly model it. Ontologies are a mechanism for representing such formal and shared domain descriptions. They can be used to annotate data with labels (metadata) indicating their meaning, thereby making their semantics explicit and machine-accessible. Many Semantic Web initiatives emphasize the capability of machines to exchange the meaning of information. Although their efforts will lead to an increased quality of the application's results, their user interfaces often take little or no advantage of the increased semantics. For example, an ontology-based search engine could use its ontology to enrich the presentation of the resulting list to the end user, for example, by replacing the endless list of hits with a navigation structure based on the semantics of the hits. Visualization is becoming increasingly important in Semantic Web tools. In par-ticular, visualization is used in tools that support the development of ontologies, such as ontology extraction tools (OntoLift, Text-to-Onto) or ontology editors (Protégé, OntoLift). The intended users of these tools are ontology engineers that need to gain an insight into the complexity of the ontology. Therefore, these tools employ schema visualization techniques that primarily focus on the structure of the ontology (i.e., its concepts and their relationships). We presented a detailed overview of these tools in Fluit et al. (2003). The Cluster Map visualization technique, developed by the Dutch software vendor Aduna (http://aduna.biz), bridges the gap between complex semantic structures and 45 46 Visualizing the Semantic Web their simple, intuitive user-oriented presentation. It presents semantic data to end users who want to leverage the benefits of Semantic Web technology without being burdened with the complexity of the underlying metadata. For end users, instance information is often more important than the structure of the ontology that is used to describe these instances. Accordingly, the Cluster Map technique focuses on visualizing instances and their classifications according to the concepts of the ontology. We have reported in previous work (Fluit et al., 2002; 2003) on case studies that exploit the expressive power of this technique. Since then, the growth of the Semantic Web has made it possible to take this technology a step further and integrate it in three different applications. Two of them are employed within Semantic Web research projects. The third is a commercial information retrieval application. These appli-cations exhibit the characteristics of a typical Semantic Web tool: they provide easy (visual) access to a set of heterogeneous, distributed data sources and rely on Semantic Web encoding languages and storage facilities for the manipulation of the visualized data. This chapter is centered on the description of these three applications. First, we will explain the contents of the Cluster Map visualization and the kind of ontologies it visualizes in Section 3.2. Section 3.3 presents the three real-life applications that incorporate the visualization. These two sections lead to a discussion in Section 3.4 on how the visualization can support several user tasks, such as analysis, search, and exploration. Some considerations for future work and a summary conclude this chapter. 3.2 Cluster Map Basics
展开▼
