Dean Allemang

What Is the Semantic Web

The Web today is something of an unruly place, with a wide variety of different sources, organizations, and styles of information. Effective and creative use of search engines is something of a craft; efforts to make order from this include community efforts like social bookmarking and community encyclopedias to automated methods like statistical correlations and fuzzy similarity matches. For the Semantic Web, which operates at the finer level of individual statements about data, the situation is even wilder. With a human in the loop, contradictions and inconsistencies in the document Web can be dealt with by the process of human observation and application of common sense. If the document Web is unruly, then surely the Semantic Web is a jungle—a rich mass of interconnected information, without any roadmap, index, or guidance. How such a mess can become something useful is the challenge that faces the working ontologist. Their medium is the distributed web of data; their tools are the Semantic Web languages RDF (Resource Description Framework), RDFS (RDF Schema), and OWL (Web Ontology Language). Their craft is to make sensible, usable, and durable information resources from this medium. The main idea of the Semantic Web is to support a distributed Web at the level of the data rather than at the level of the presentation. Instead of having one webpage point to another, one data item can point to another, using global references called uniform resource identifiers (URIs). The Web infrastructure provides a data model whereby information about a single entity can be distributed over the Web.

Chapter 3 – RDF—The basis of the Semantic Web

Publisher Summary RDF is a system for modeling data. It gives up in compactness what it gains in flexibility. Every relationship between any two data elements is explicitly represented, allowing for a very simple model of merging data. A relationship is either present or it is not. Merging data is thus reduced to a simple matter of considering all such statements from all sources, together in a single place. The only challenge that remains in such a system is the challenge of identity. This problem is not unique to the RDF data model. The infrastructure of the Web itself has the same issue and has a standard solution: the URI. RDF borrows this solution. Since RDF is a Web language, a fundamental consideration is the distribution of information from multiple sources across the Web. On the Web, the AAA slogan holds: Anyone can say Anything about Any topic. RDF supports this slogan by allowing any data source to refer to resources in any namespace. Even a single triple can refer to resources in multiple namespaces. As a data model, RDF provides a clear specification of what has to happen to merge information from multiple sources.

Chapter 2 – Semantic Modeling

Publisher Summary 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. The modeling activity is the activity of distilling communal knowledge out of a chaotic mess of information. Human communication as a goal for modeling allows it to play a role in the ongoing collection of human knowledge. The levels of communication can be quite sophisticated, including the collection of information used to interpret other information. In this sense, human communication is the fundamental requirement for building a Semantic Web. It allows people to contribute to a growing body of knowledge and then draw from it. But communication is not enough; to empower a web of human knowledge, the information in a model needs to be organized in such a way that it can be useful to a wide range of consumers. Models are used to organize human thought in the form of explanations.

Semantic Web application architecture

Many of the components of a Semantic Web application are provided as supported products by companies specializing in Semantic Web technology or by free software under a variety of licenses. New software is being developed by research groups as well as product companies on an ongoing basis. This chapter describes types of components that make up a Semantic Web deployment and how they fit together. The components such as RDF (Resource Description Framework) parsers, serializers, stores, and query engines are not semantic models in themselves but the components of a system that will include semantic models. Even the information represented in RDF is not necessarily a semantic model. These are the building blocks that go into making and using a semantic model. The model will be represented in RDF, to be sure. A semantic model acts as a sort of glue between disparate, federated data sources so one can describe how they fit together. Just as Anyone can say Anything about Any topic, so also can anyone say anything about a model; that is, anyone can contribute to the definition and mapping between information sources. In this way, not only can a federated, RDF-based, semantic application get its information from multiple sources, but it can also get the instructions on how to combine information from multiple sources. In this way, the Semantic Web really is a web of meaning, with multiple sources describing what the information on the Web means.

Ontologies on the Web—putting it all together

This chapter discusses three ontologies: Good Relations, Quantities/Units/Dimensions/Types (QUDT), and OBO Foundry. These ontologies cover the spectrum from ontologies that include almost no data at all (Good Relations) to ontologies that include very large amounts of richly interconnected data (OBO). They all supply the basic capabilities of the Semantic Web by sharing information in a coherent way across multiple systems. Good Relations is the smallest of the ontologies described in the chapter. Its main goal in the Semantic Web is to provide a framework in which information can be shared—a vocabulary that different suppliers can use to describe their offerings. The data in Good Relations aren’t in the ontology at all; it is distributed across the Web. OBO Foundry ontologies, in contrast, are larger, and include large amounts of data about biology, medicine, life sciences, etc. There are complex questions that can be answered using OBO as a data resource. QUDT sits in the middle; it contains a good deal of data, but its main purpose is to provide connection between other data sets; two data sources that both use Good Relations might still fail to be interoperable because of mismatch of units; QUDT provides enhanced interoperability in these cases. All three of these ontologies play the basic role in the Semantic Web of providing globally unambiguous names for standard entities—they differ only in the details of how these relationships can be used.

Chapter 10 – SKOS—managing vocabularies with RDFS-Plus

SKOS (Simple Knowledge Organization System) is a W3C Recommendation that provides a means for representing knowledge organization systems in a distributed and linkable way. SKOS was designed from the start to allow modelers to create modular knowledge organizations that can be reused and referenced across the Web. It was designed to augment thesaurus standard by bringing the distributed nature of the Semantic Web to thesauri and controlled vocabularies. Toward this end, it was also a design goal of SKOS that it be possible to map any thesaurus standards to SKOS in a fairly straightforward way. SKOS takes advantage of the distributed nature of RDF to allow extension to a network of information to be distributed across the Web. It relies on the inferencing structure of RDFS-Plus to add completeness to their information structure. SKOS vocabularies provide a cornerstone for linking information on the Web. Publishing vocabularies in SKOS allows the concepts they define to be referenced on a global scale.

Querying the Semantic Web—SPARQL

RDF provides a simple way to represent distributed data. The triple is the simplest way to represent a named connection between two things. But a representation of data is useless without some means of accessing that data. The standard way to access RDF data uses a query language called SPARQL. SPARQL stands for SPARQL Protocol And RDF Query Language. The SPARQL query language works closely with the structure of RDF itself. SPARQL query patterns are represented in a variant of Turtle. The queried RDF graph can be created from one kind of data or merged from many; in either case, SPARQL is the way to query it. This chapter gives examples of the SPARQL query language. Most of the examples are based on version 1.0 of the standard. SPARQL is a query language and shares many features with other query languages like XQUERY and SQL. But it differs from each of these query languages in important ways. SPARQL queries can be used to fetch information (like SQL queries) or to transform a graph into a new form (like rules). Both forms use the same notion of graph pattern to specify the desired information.

Chapter 11 – Basic OWL

Publisher Summary This chapter presents the modeling capabilities of Web Ontology Language (OWL) that go beyond RDFS-Plus. A key functionality of OWL is the ability to define restriction classes. The unnamed classes are defined based on restrictions on the values for particular properties of the class. Using this mechanism, OWL can be used to model situations in which the members of a particular class must have certain properties. In RDFS, the domain and range restrictions can allow one to make inferences about all the members of a class (such as plays relating a baseball player to a team). In OWL, one can use restriction statements to differentiate the case between something that applies to all members of a class versus some members, and even to insist on a particular value for a specific property of all members of a class. When restrictions are used in combination with the constructs of RDFS, and then they are cascaded with one another, they can be used to model complex relationships between properties, classes, and individuals. The advantage of modeling relationships in this way (over informal specification) is that interactions of multiple specifications can be understood and even processed automatically.

Chapter 9 – Using RDFS-Plus in the wild

Publisher Summary This chapter describes two example uses of the RDFS-Plus constructs. Both of these applications of RDFS-Plus make essential use of the constructs in RDFS-Plus, though often in quite different ways. These are real modeling applications built by groups who originally had no technology commitment to RDFS or OWL (though both were conceived as RDF applications). In both cases, the projects are about setting up an infrastructure for a particular Web community. The use of RDFS-Plus appears in the models that describe data in these communities, rather than in the everyday use in these communities. This chapter describes how modeling works in RDFS and OWL; hence the focus is on the community infrastructure of these projects. The first application is part of a major US government effort called Data.gov. Data.gov is an effort made by the US government to publish public information. There are hundreds of thousands of datasets in Data.gov, of which hundreds are made available in RDF, with many more being converted all the time. Data.gov is a great example of the data wilderness; the published data sets come from a wide variety of source formats and collection methodologies, resulting in idiosyncratic data representations. The second application is called FOAF, for “Friend of a Friend.” FOAF is a project dedicated to creating and using machine-readable homepages that describe people, the links between them, and the things they create and do. FOAF was originally based on RDF because of the inherently distributed and Web-like nature of the project requirements. As the project evolved, there was a need to describe the relationships between various resources in a formal way; this led it to RDFS and then on to RDFS-Plus. The chapter describes each of these efforts and shows their use of the RDFS-Plus.

Chapter 12 – Counting and sets in OWL

Publisher Summary The restrictions are powerful methods for defining classes of individuals. This chapter helps see that Web Ontology Language (OWL) augments this capability with a full set theory language, including intersections, unions, and complements. These can be used to combine restrictions together (e.g., the set of planets that go around the sun and have at least one moon) or to combine the classes one uses to define restrictions (a vegetarian is someone who eats food that is not meat). This combination provides a potent system for making very detailed descriptions of information. OWL also includes restrictions that refer to cardinalities—that is, referring to the number of distinct values for a particular property some individual has. Reasoning with cardinalities in OWL is surprisingly subtle. Perhaps one shouldn’t be surprised that it is difficult to count how many distinct things there are when one thing might have more than one name (i.e., more than one URI), and one never knows when someone might tell about a new thing one didn’t know about before. These are the main reasons why cardinality inferencing in OWL is quite conservative in the conclusions it can draw.