Article Preview
Top1. Introduction
The goal of the Semantic Web is to enable machines to more easily exchange, merge and reuse datasets by creating a Web of data where the data has clearly-defined, machine-processable semantics (Shadbolt, Berners-Lee, & Hall, 2006). The languages of the Semantic Web are designed to facilitate machine interpretability of information and do not define a visual presentation model since human readability is not among their primary goals. However, these data are, at least partially, created and manipulated by people: domain experts who do not necessarily have advanced skills in computer science but still have to understand the data; software developers that write programs to query and manipulate the data. Downward in the data processing pipeline, end-users eventually make use of the data, or a subset of it, presented to them in some form or another (Dix et al., 2010). In any of the above cases, the data have to be displayed in a human-friendly way, that raw textual serializations do not allow, requiring solutions for data restructuring and transformation. Targets of these transformations can be documents such as HTML pages, interactive graphical components such as maps and bar charts, or advanced information visualization components such as those developed for graph visualization (Herman, Melancon, & Marshall, 2000).
As Semantic Web technologies are slowly but steadily gaining adoption, more and more datasets are made available in the form of populated ontologies, many of which contain millions of triples. Significant research and development effort has been dedicated to the design of visualization tools for Semantic Web data (Katifori, Halatsis, Lepouras, Vassilakis, & Giannopoulou, 2007), ranging from visualization plugins for integrated development environments such as Protégé (Knublauch, Fergerson, Noy, & Musen, 2004) to more exotic, virtual-reality based visualizations that run in immersive environments (Halpin, Zielinski, Brady, & Kelly, 2008). By making use of the richer capabilities of graphical representations, as opposed to textual representations such as N3 or RDF/XML, and by abstracting from the complex syntactic details of the latter and explicitly representing relations, visual tools aim at providing better cognitive support to users (Ernst, Storey, & Allen, 2005), from knowledge engineers to domain-expert end-users. They provide them with interactive representations of the data based upon state-of-the art information visualization techniques, better supporting tasks such as ontology understanding, discovery, search (Ernst et al., 2005), comparison and mapping (Lanzenberger & Sampson, 2006; Falconer & Storey, 2007).
Most tools structure the visualization according to the data model, i.e., to the concept definitions and interrelations that constitute the ontology's vocabulary (the TBox in Description Logics). While many tools do support the visualization of instance (ABox) data, instances are often treated as somewhat peripheral information. The visualization is mainly structured according to the TBox, instances that constitute the ABox being treated as leaf nodes in this tree or graph structure. Several exceptions to this general observation exist, but either give a limited view of the ABox (Fluit, van Harmelen, & Sabou, 2003; Tu et al., 2005; Lohmann, Heim, Stegemann, & Ziegler, 2010; Motta et al., 2011) or use conventional node-link diagram representations that hardly scale beyond a few hundred nodes at best (Noppens & Liebig, 2006; Kriglstein & Wallner, 2010).
Yet, instances, that populate ontologies, represent an essential part of the overall knowledge base. Understanding instance-level data might be easier for users because of their lower level of abstraction compared to the definition of concepts based on OWL constructs, but instances will often be orders of magnitude more numerous than the definitions that give them machine-processable meaning (see, e.g., many of the datasets currently part of the Linking Open Data graph1). As such, the visualization of instance-level data poses different but real challenges that remain to be addressed.