This paper responds to a set of general questions that might be asked of any "emerging technology" in the realm of XML. The questions of utility, viability, and the technical implications of adopting any or these standards calls for careful scrutiny and the need to keep up to date on progress. I've chosen RDF because of an inherent interest in metadata especially as applied to web resources.

I have elected in fact to create a set of RDF metadata about this very paper, its author (me), the course and section I am in, the teacher of the course, and the teaching fellow who will correct the paper. These RDF graphs and "triples" (constructed using the RDF Validator) are noted below towards the end of the paper.

Finally, as another "hands-on" XML twist, I decided to write the paper not in Microsoft Word, but in DocBook. This has been a very good learning experience as well. I have used the DocBook "Simplified DocBook" DTD for XML, and even its vocabulary of elements takes some getting familiar with (106 elements, 525 entities, and 26 notations). As for DocBook XSLT stylesheets, I've simply used the basic XHTML version, though I'd like to take this further (later) and learn how to create a .PDF version of it, and to "chunk" a single document into separate, navigable HTML pages, etc.

Despite what one might intuit (based perhaps on familiarity with <META> tags in HTML files), the RDF fragment itself is not included as part of this XML DocBook document. There are a couple of reasons for this. First, DocBook, being a DTD (though a XML Schema version ca. 2000-2001 was beta-released) cannot handle namespaces (well), and the RDF specification uses them heavily (in fact RDF was the first specification to introduce the real need for namespaces to the XML family of technologies).

The second reason is perhaps more interesting: the concept of Semantic Web RDF metadata is that it shouldn't have to be stored in or with the resources it describes. In many cases people will be creating metadata about resources they have no write access to.

Finally, this set of RDF, despite being very small, is in most part not really about this final paper document. It is instead about writers and correctors of the document, and their relationships by means of courses and sections, etc. So the insertion of this RDF into this document would not really be relevant or appropriate. Metadata in the Semantic Web is going to have a life and place of its own, beyond and apart from the web information resources it describes.

The Resource Description Framework (RDF), at a high level, is about the metadata that will be needed to “[lead] the Web to its full potential,” as stated on the homepage of the World Wide Web Consortium. RDF is the centerpiece of what is arguably the key initiative at the W3C these days, “ The Semantic Web". This has been defined by Tim Berners-Lee as "an extension of the current web, in which information is given well-defined meaning, better enabling computers and people to work in cooperation.” The RDF work at the W3C released its first working draft in October, 1997, more or less coincident with the beginning of the "Semantic Web Activity, in the W3C's "Technology and Society" domain.

Other groups have been active in this arena, and some are skeptical of the quality or the likelihood of success of this W3C initiative. The IEEE has its IEEE P1600.1 Standard Upper Ontology (SUO) Working Group and 'KIF'. According to a writer from SUO (at http://perso.club-internet.fr/jld/billquote.html), RDF notation was originally designed by R. V. Guha, former associate director of Cyc. He states that RDF "is a trivial subset of what Cyc was doing in 1984." Other concerns have been expressed as well. On Slashdot a year ago (21 March 2001) "Is The Semantic Web A Pipe Dream?" appeared with several points made about the difficulty of getting enough people onboard with this to create the necessary network effects to make it really work.

Setting aside these concerns for the moment, to give some sense of concrete application to the W3C claims for the promise of the Semantic Web, the following (quite!) futuristic scenario appeared in the Scientific American article on the subject of the Semantic Web last year (by Tim Berners-Lee and others):

http://www.sciam.com/article.cfm?articleID=00048144-10D2-1C70-84A9809EC588EF21&ref=sciam

“ The entertainment system was belting out the Beatles' "We Can Work It Out" when the phone rang. When Pete answered, his phone turned the sound down by sending a message to all the other local devices that had a volume control . His sister, Lucy, was on the line from the doctor's office: "Mom needs to see a specialist and then has to have a series of physical therapy sessions. Biweekly or something. I'm going to have my agent set up the appointments." Pete immediately agreed to share the chauffeuring. At the doctor's office, Lucy instructed her Semantic Web agent through her handheld Web browser. The agent promptly retrieved information about Mom's prescribed treatment from the doctor's agent, looked up several lists of providers , and checked for the ones in-plan for Mom's insurance within a 20-mile radius of her home and with a rating of excellent or very good on trusted rating services. It then began trying to find a match between available appointment times (supplied by the agents of individual providers through their Web sites) and Pete's and Lucy's busy schedules. (The emphasized keywords indicate terms whose semantics, or meaning, were defined for the agent through the Semantic Web.)...”

Or perhaps a little less breathlessly, again Tim Berners-Lee (http://www.w3.org/DesignIssues/Business) (Oct. 2001) "We're working to the time when you can click on the web page for the meeting, and your computer, knowing that it is indeed a form of appointment, will pick up all the right information, and understand it enough to send it to all the right applications"

In sum, the problem RDF sets out to solve is that the mass of information on the web has no metadata that would permit systems to interoperate and discover resources that match certain requirements, based on content and subject matter. RDF (and its attendant technologies, including “DAML+OIL” (DARPA Agent Markup Language + Ontology Inference Layer, or Ontology Interchange Language), XTM (Topic Maps), and Web Ontologies) seeks to address this lack of metadata on the Web today. Or, to put it another way, RDF seeks to be the solution to providing a framework for the representation of the semantic meaning of web resources, such that those resources will be able to be utilized more fully, and especially by automated agents. While it's true that today documents can be created in the Extensible Markup Language (XML) to ensure the structure and even data typing (with XML Schema) are controlled and valid, the actual meaning, semantics, or significance of any particular data is not automatically available to other systems by means of what XML can offer.

As an example, if my music collection document uses <author>Mozart</author> and your music collection document reads <composer>Mozart</composer>, then only by unacceptably manual mapping will be be able to share information resources. (Or as another example, all those healthcare providers in the scenario above would have to use some kind of machine-understandable and standard markup tagging to indicate “location,” so the web agent would be able to calculate which ones were within a 20-mile radius.)

The Resource Description Framework provides the means to use unique identifiers (URIs) to unambiguously indicate metadata information about a resource. The adoption of shared unique identifiers by multiple parties (e.g. the classic example ISO country or state codes, but also industry-standard codes, or even narrower, proprietary identifiers like a college’s student list) provides the way for automated systems to make connections among web resources, some of which might be explicitly defined, while others remain unconnected explicitly but are derivable by inference. The potential is great for applications today unforeseeable, as we have witnessed with so much innovation within the Web to date, only the next wave of innovation will have much to do with automated agents pulling information from the web, rather than human agents going around to find and retrieve the information.

The first RDF draft at the W3C was published in October 1997, and it has had therefore a comparatively long history and a wide degree of support and interest at the W3C. It thus far is mostly a technology being embraced by universities, the W3C, and the open developer community, and does not appear to factor into publicized plans or certainly not into marketing materials from large corporations like Microsoft, Oracle, etc., the way some other W3C initiatives have (e.g. XML itself of course). The list of members of the Core RDF Working Group (http://www.w3.org/2001/sw/RDFCore/members.html) does include Nokia, Hewlett-Packard, and I.B.M.

This work is not immediately seen as being so much about commerce (though applications will certainly be there for it), but more about interoperability and information exchange, which at times is not what commerce seeks, instead preferring proprietary and closed information, such that potential customers “stick” to the site or company they are at, and do their purchasing there, instead of being freely interconnected to competing possibilities in the wide world of the web.

As stated on a slide from a presentation to be given by Ora Lassila this month (May 2002) at the 11th WWW Conference in Hawaii http://www.lassila.org/publications/2002/lassila-www2002-rdf-panel.pdf, the viability of this Semantic Web ideal will have to have some real business model underpinnings:

So if users aren't looking at screens and encountering advertising, will they tolerate advertising mixed in with their agent report results? Or will people pay for agent reports in a way that we are not used to today re: getting information from the web? Remains to be seen.

On another topic, the lack of technology and applications for this RDF metadata framework is a cause for concern, after all this time. Many complain that it is too difficult to work with (recent xml.com article on the new RDF Primer contains the subtitle “The real problem with RDF” (http://www.xml.com/pub/a/2002/0/15/rdfprimer.html). Still, there is a great deal of activity and some real projects employing it (Dublin Core and others), so it unquestionably merits further close attention. But to temper that, the “Grand Unifying Theory” nature of it all (to quote from another article on it) can be overwhelming and make it appear simply too ambitious.

These two diagrams help understand the key position of RDF to the higher schema and ontology layers that are based on its fundamental data model of the triple. They also help visualize the relationship that the XML syntax plays, as an implementation choice for the RDF model, but the diagrams show how RDF does provide a layer on top of what XML can do vis-a-vis metadata semantic and meaning.

Ora Lassila: (http://www.lassila.org/publications/) "The Resource Description Framework", IEEE Intelligent Systems 15(6): 67-69 (November/December 2000)

http://www.computer.org/intelligent/ex2000/pdf/x6067.pdf

Tim Berners-Lee http://www.w3.org/2000/Talks/1206-xml2k-tbl/slide10-0.html

There are issues that RDF alone cannot handle, which brings up the topic of the RDF Schema (RDFS). Essentially, this ensures 1) that the Properties assigned to a Resource make sense (you can't have "Date Of Birth" for the name of a state county), and 2) that the Values assigned to a Property make sense (you can't have "New York, New York" for a "Date Of Birth" property). The RDFS is outside the scope of this short paper. Another layer on top of RDFS is the ontology layer of DAML+OIL (DARPA Agent Markup Language + Onotology Inference Layer (or Interchange Language), both of which also must be outside the scope of this paper. But the thing to realize is that these higher level layers are all dependent upon the essential RDF data model, namely the triple.

In each of the cases just described, the technologies, tools, trained developers, and really even the specifications are not actually yet in place to do a lot of these things. There are examples of RDF in practice, but some or most of these examples are fairly close to the original core group of specification creators (e.g. The Dublin Core Metadata Initiativeuses it, and Dublin Core is one of the key vocabularies used and endorsed for RDF). The publishing industry has since 1999 been developing PRISM (Publishing Requirements for Industry Standard Metadata). "PRISM is an extensible XML metadata standard for syndicating, aggregating, post-processing and multi-purposing content from magazines, news, catalogs, books and mainstream journals."

The RDF model it appears is a solid base from which to begin to create additional layers of metadata modelling and representation, adding features and capabilities. The "triples" are the key to this solid base. On top of that the prevalent ideas in force are the DARPA Agent Markup Language, paired with the Ontology Inference Layer (DAML+OIL). A related area of investigation is the XTM Topic Maps, which differs from RDF but has much in common with it as well. Work is underway to see what kind of congruence there could be between these varying metadata approaches.

Edd Dumbbill of xml.com wrote a useful assessment a year ago (http://www.xml.com/pub/a/2001/03/07/buildingsw.html) in which he noted that "the builders of the Semantic Web [should] keep their feet on the ground. ... Rather than pondering grand unification theories, we should concentrate on doing small things well and solving achievable and well-defined problems "

It is difficult to know what to conclude from this brief look into the state of affairs of this major initiative, which claims no less than being the basis for the "2nd wave of the World Wide Web." A great deal of thinking, energy, and intelligence is going into the basic specification work. Next remains to be seen just what tools will be developed, and what business models will be made to work to support the investment in the metadata creation, to achieve the sought-after vision of a truly interoperable Web, made possible by machine-understandable metadata.