is a tradeoff when we model, and although Anyone can say Anything about Any topic, not everyone will want to say certain things. There are those who are interested in saying details about individual entities, like the preferred symbol for Pluto or the themes in life that it signifies. Others (like the IAU) are interested in talking about categories, what belongs in a category, and how you can tell the difference. Still others (like lexicographers, information architects, and librarians) want to talk about the rules for specifying information, such as whether there can be more than one preferred label for any entity. All of these people have contributions to make to the web of knowledge, but the kinds of contributions they make are very different, and they need different tools. This difference is one of level of expressivity.
The idea of different levels of expressivity is as well known in the history of collaborative human knowledge as modeling itself. Take as an example the development of models of a water molecule, as shown in Figure 2.4. In part (a), we see a model of the water molecule in terms of the elements that make up the molecule and how many of each is present—namely, two hydrogen atoms and one oxygen atom. This model expresses important information about the molecule, and it can be used to answer a number of basic questions about water, such as calculating the mass of the molecule (given the masses of its component atoms) and what components would have to be present to be able to construct water from constituent parts.
In Figure 2.4(b), we see a model with more expressivity. Not only does this model identify the components of water and their proportions, but it also shows how they are connected in the chemical structure of the molecule. The oxygen molecule is connected to each of the hydrogen molecules, which are not (directly) connected to one another at all. This model is somewhat more expressive than the model in part (a); it can answer further questions about the molecule. From (b), it is clear that when the water molecule breaks down into smaller molecules, it can break into single hydrogen atoms (H) or into oxygen-hydrogen ions (OH) but not into double-hydrogen atoms (H2) without some recombination of components after the initial decomposition.
Finally, the model shown in Figure 2.4(c) is more expressive still in that it shows not only the chemical structure of the molecule but also the physical structure. The fact that the oxygen atom is somewhat larger than the hydrogen atoms is shown in this model. Even the angle between the two hydrogen atoms as bound to the oxygen atom is shown. This information is useful for working out the geometry of combinations of water molecules, as is the case, for instance, in the crystalline structure of ice.
Just because one model is more expressive than another does not make it superior; different expressive modeling frameworks are different tools for different purposes. The chemical formula for water is simpler to determine than the more expressive, but more complex, models, and it is useful for resolving a wide variety of questions about chemistry. In fact, most chemistry textbooks go for quite a while working only from the chemical formulas without having to resort to more structural models until the course covers advanced topics.
Figure 2.4 Different expressivity of models of a water molecule.
The Semantic Web provides a number of modeling languages that differ in their level of expressivity; that is, they constitute different tools that allow different people to express different sorts of information. In the rest of this book, we will cover these modeling languages in detail. The Semantic Web standards are organized so that each language level builds on the one before so the languages themselves are layered. The following are the languages of the Semantic Web from least expressive to most expressive.
RDF—The Resource Description Framework. This is the basic framework that the rest of the Semantic Web is based on. RDF provides a mechanism for allowing anyone to make a basic statement about anything and layering these statements into a single model. Figure 2.3 shows the basic capability of merging models in RDF. The work on RDF started in 1997 and it has been a recommendation from the World Wide Web Consortium (W3C) since 1999, updated in 2004 and in 2014 with RDF 1.1.
SHACL—The Shapes Constraint Language. SHACL is a language based on the intuition that we expect data to be in a certain form, or shape. SHACL allows a modeler to represent the expected shape of a data description. These shapes can be used to validate data or to present a form to a human user to fill out to supply data. Unlike the other Semantic Web modeling languages, which are designed based on the Open World Assumption, SHACL works with the Closed World Assumption; if data is not included in a description, then it is considered to be missing. SHACL is one of the newest modeling languages in the Semantic Web stack, and became a W3C Recommendation in 2017.
RDFS—The RDF Schema Language. RDFS is a language with the expressivity to describe the basic notions of commonality and variability familiar from object languages and other class systems—namely classes, subclasses, and properties. Figures 2.1 and 2.2 illustrate the capabilities of RDFS. RDFS was drafted in 1999 and became a W3C Recommendation in 2004.
RDFS-Plus. RDFS-Plus is a subset of Web Ontology Language (OWL) that is more expressive than RDFS but without the complexity of OWL. There is no standard in progress for RDFS-Plus, but there is a growing awareness that something between RDFS and OWL could be industrially relevant. We have selected a particular subset of OWL functionality to present the capabilities of OWL incrementally. RDFS-Plus includes enough expressivity to describe how certain properties can be used and how they relate to one another. RDFS-Plus is expressive enough to show the utility of certain constructs beyond RDFS, but it lacks the complexity that makes OWL daunting to many beginning modelers. The issue of uniqueness of the preferred symbol is an example of the expressivity of RDFS-Plus.
OWL—the Web Ontology Language. OWL brings the expressivity of logic to the Semantic Web. It allows modelers to express detailed constraints between classes, entities, and properties. OWL was adopted as a recommendation by the W3C in 2004, with a second version adopted in 2009.
The W3C provides a number of standards built on this stack to manage provenance, services, data catalogs, online analytical processing (OLAP), and a variety of other things, many of which we will treat in later chapters. But the languages listed here are the foundational modeling languages that the others build on, and are the main topic of this book.
2.5 Summary
The Semantic Web, just like the hypertext Web that preceded it, is based on some radical notions of information sharing. These ideas—the AAA slogan, the Open World Assumption, and nonunique naming—provide for an environment in which information sharing can thrive and a network effect of knowledge synergy is possible. But this style of information gathering creates a chaotic landscape rife with confusion, disagreement, and conflict. How can the infrastructure of the Web support the development from this chaotic state to one characterized by information sharing, cooperation, and collaboration?
The answer to this question lies in modeling. Modeling is the process of organizing information for community use. Modeling supports this in three ways: It provides a framework for human communication, it provides a means for explaining conclusions, and it provides a structure for managing varying viewpoints. In the context of the Semantic Web, modeling is an ongoing process. At any point in time, some knowledge will be well structured and understood, and these structures can be represented in the Semantic Web modeling language. At the same time, other knowledge will still be in the chaotic, discordant stage, where everyone is expressing himself differently. And typically, as different people provide their own opinions about any topic under the sun, the Web will simultaneously contain organized and unorganized knowledge about the very same topic.
