The Semantic Web: A Path to Large-Scale Interoperability

By Frank Manola, Mary Pulvermacher, and Leo Obrst

he World Wide Web is an extreme example of an environment where a large-scale collection of dissimilar computer systems needs to interoperate and, thus, provides a "grand challenge" for any approach to interoperability. The Semantic Web aims to carry this interoperability to the next level: beyond today's Web of human-readable Web pages to a global database of "machine-readable" and "machine-interpretable" information.

Today's Web is oriented toward presentation to people: this is the purpose of html tags <bold>PRINT THIS IN BOLD</bold> on Web pages. After HTML came Extensible Markup Language (XML), which separates the presentation from the meaning <LastName>Obrst</LastName>. However, to a computer this is the same as <%@*&^>Obrst</%@*&^>. A machine needs something that helps interpret what the tags mean. That's were "machine-interpretable" semantics comes in.

The Semantic Web is a more efficient way of representing data on the Web. A major thrust of this effort is developing languages for explicitly describing the "meaning" of Web resources, both information and services, so that they can be "understood" by software that needs to select among them or combine them in appropriate ways.

Development is being done by an international Semantic Web research activity, spearheaded by the World Wide Web Consortium (W3C) and the Defense Advanced Projects Research Agency (DARPA) Agent Markup Language (DAML) program. Semantic Web technologies are being applied not only to improve interoperability on the open Web, but also to improve information and service interoperability in more localized environments, such as within intranets protected by firewalls from the full outside Internet.

MITRE is influencing the emergence of the Semantic Web in a number of ways: both in the standards arena and in research. MITRE has been participating in developing standards for both the W3C Resource Description Framework (RDF) and the RDF-based ontology language called OWL, and we are beginning to participate in the emerging Semantic Rules, Semantic Web Services, and Semantic Web/Ontologies Lessons Learned Working Groups, which are just getting started. In addition, we are performing research related to the Semantic Web in the areas of semantic data integration and interoperability, characterizing the semantics of messages—as in United States Message Text Format (USMTF). We are also involved in sponsors' projects that are capitalizing on the Semantic Web technologies, especially in intelligence programs and in emerging efforts, such as the federal e-Gov and Department of Defense net-centric operations initiatives.

A key part of the Semantic Web is the idea of describing things using simple subject-predicate-object statements or triples, such as serial number 82735 is an Aircraft, or serial number 82735 needs part part456 using RDF. RDF provides an XML-based language for writing such statements, using vocabularies based on Web Uniform Resource Identifiers (URIs) rather than on ordinary words. For example, in writing these statements in RDF, the URI http://usaf.example.mil/inventory#sn82735 might be used to identify the individual aircraft (this URI could be abbreviated as usaf:sn82735, using XML's namespace mechanism). Hence, corresponding RDF statements could be written as:
usaf:sn82735 rdf:type usaf:Aircraft
usaf:sn82735 usaf:needsPart usaf:part456

Using URIs allows RDF to describe things on the Web, even when the things (like an aircraft) can't be retrieved on the Web. Using URIs instead of words also allows different users or domain communities to develop their own vocabularies, with distinct meanings assigned to what otherwise might be thought to be the same term (such as "aircraft"). In addition, further information about the identified item can be stored at the indicated Web URI and accessed through the Web. Collections of RDF statements form graphs or webs of relationships among the terms used, and data in other XML languages, relational databases, and many other formats can be readily interpreted as RDF statements.

RDF's uniform triple structure eliminates one major problem in integrating data from different sources: having to deal with the different structures involved. With RDF, you simply merge the collections of triples. For example, RDF's simple data structure, together with a simple shared vocabulary, allow applications using RDF Site Summary, a widely-used Web language for simple content syndication, to easily merge news items from multiple sources all over the Web into a common presentation. RDF allows terms from multiple vocabularies to be freely mixed.

In addition to providing RDF to describe things using simple statements, the Semantic Web effort has given us the RDF-based ontology language called OWL. An ontology provides a machine-processable description of the terms (such as usaf:Aircraft or usaf:needsPart) that an organization or application can use, as well as aspects of the meanings of those terms. When two or more organizations or applications need to interoperate, their ontologies can provide the basis for understanding the terms each is using, the differences among them, and how to resolve those differences.

For example, an OWL ontology might specify that an F15 is a kind of aircraft using the statement:
usaf:F15 rdfs:subClassOf usaf:Aircraft
so that, based on this declaration, an F15 could be properly interpreted, at least to some extent, by a program that doesn't know about F15s, but does know about aircraft. Because OWL is based on RDF, an OWL ontology simply enriches the original graph of information by adding further triples to it.

Beyond ontologies, the Semantic Web defines a rule layer to further enrich these descriptions, e.g., to define myOnt:altitude as distance over the earth's surface (an ontology used by another organization might define anotherOnt:altitude as distance from the earth's center), or to state that myOnt:weight in kilograms = myOnt:weight * 2.2046 in pounds.

Because these terms are URIs rather than simple words, their definitions and related information could be directly accessed on the Web. As a result, it becomes easier to discover, create, and relate information. Moreover, within the common environment of the Web, people and software can begin to discover and use common references to common information, creating a rich Semantic Web of shared information and meaning.

Web services are tied into this framework using a specialized ontology that describes the services more richly than conventional Web service description languages. The DARPA Agent Markup Language (DAML) program has developed one such ontology, called DAML-Services (DAML-S), which has evolved to OWL-S. This is an active research effort to develop a Web service ontology to use to describe the properties and capabilities of Web services in unambiguous, computer-interpretable form.

Of course, large scale interoperability raises a host of other issues, such as the trustworthiness of data obtained from multiple sources. Further technical layers of the Semantic Web are being developed to address such issues.

In a number of MITRE projects, we are both developing and applying Semantic Web technologies. Several MITRE employees are directly participating in developing the key W3C Semantic Web languages, RDF and OWL. In addition, we are applying Semantic Web technologies to customer applications in areas such as intelligence analysis and command and control. The DAML program is also funding development of both prototype applications and tools in these and other areas.

While the word "semantic" in "Semantic Web" may suggest an air of unreality, some of the Semantic Web technologies are in use today, addressing practical interoperability problems in such areas as bioinformatics, real estate, publishing, electric power control, intelligence, planning, and command and control. And its use is increasing. Moreover, commercial vendors are now supporting RDF and OWL with new tools.

Research Efforts

Acknowledging the growing interest in and the promise of Semantic Web technologies, MITRE invested its own research funds by awarding the Semantic Web Innovation Grant in November 2003. This award encourages MITRE researchers to respond to the challenges of the Semantic Web, which will affect all of our sponsors. The results of the Innovation Grant team's investigation on the Semantic Web's maturity and how related technologies can be applied to benefit our customers in the execution of their missions is being shared throughout MITRE.

Conclusion

Semantic Web technologies will not take over the Web all at once. Instead, different application domains will determine the appropriateness of using these technologies using their own criteria, their own timeline, and their own evaluation of cost-benefit tradeoffs. However, the fact that some application domains are making use of these technologies right now suggests that the answer to "when" is sooner rather than later. Supported by the base W3C standards, which are nearly completed, and numerous prototypes and operational applications, the Semantic Web is poised to take off.

For more information, please contact Leo Obrst or Mary Pulvermacher using the employee directory.

Solutions That Make a Difference.®
Copyright © 1997-2013, The MITRE Corporation. All rights reserved.
MITRE is a registered trademark of The MITRE Corporation.
Material on this site may be copied and distributed with permission only.

Privacy Policy | Contact Us