OK, so you want to participate in the "second Web revolution", but need more inside information? This concept has already created a high degree of disagreement, particularly relating to its scope. However, the W3C states that the Semantic Web Activity is a successor to the W3C Metadata Activity, so your previous knowledge in this area (Dublin Core Metadata etc.) will be very useful. Most of the metatags used before were proprietary and had no contextual meaning in general search applications. On the contrary, semantic Web will be decentralized and based on a RDF (Resource Description Framework) which uniquely and unambiguously describes the content on a particular Web page. Although its unwieldy syntax isn't very popular between Web developers, the concepts behind this technology are rather clear and easy to understand. XML (extensible Markup Language) is often used to serialize RDF models, but the similarity stops here: XML schemata operate at the "syntax level", while RDF works at the data model level, allowing machines to make inferences about the collected data. In the simplest incarnation, RDF data consists of nodes and corresponding attribute/value pairs. Node is usually a Web resource or other instance of metadata - basically anything that can be pointed by an Universal Resource Identifier (URI). Attributes are named properties of the nodes, and their values are either primitive data types (strings, numbers) or other resources and metadata instances. This mechanism allows for building structures known as labeled directed graphs.

This model is conceptualized by using RDF "sentences" or triples, consisting of a subject, a property, and an object. Unlike the ambiguity of every natural language, this model presents knowledge in a unique and unambiguous fashion. Consider the following sentences:

• The article titled "RDF for beginners" is authored by John Doe.
• John Doe wrote an introductory paper titled "RDF for beginners".

The same fact may be expressed using many syntactical forms, although each one of them may convey the same meaning. Humans are extremely adept at extracting meaning from differing syntactic constructs, but machine will see only a few entirely different strings. Imagine the complexity of a Natural Language Processing system that could extract the meaning from the texts found in everyday newspapers or similar information sources. But RDF gives machines a chance:

<?xml:namespace ns = "http://www.w3.org/RDF/RDF/" prefix ="RDF" ?>
<?xml:namespace ns = "http://somesite.com/NS/" prefix = "NS" ?>

<RDF:RDF>
  <RDF:Description RDF:HREF = "http://somesite.com/somedoc.htm">
    <NS:Title>RDF for beginners</NS:Title>
    <NS:Author>John Doe</NS:Author>
  </RDF:Description>
</RDF:RDF>

The URI associated with the namespace declaration at the beginning of this code snippet references the corresponding XML scheme. Just apply that knowledge and voila... the semantic information from the target resource can be easily extracted.

This infrastructure gives us a lot of opportunities, but grand plans for classification of almost everything are still notoriously difficult to implement. The reason lies in the unavailability of industry-wide ontologies (hierarchies of classes and the relations between them) and tools to manage and maintain them. For example, if major manufacturers of computer parts cannot agree on common standards for the exchange of metadata, the functionality of semantic Web for this industry will be severely limited. Another challenge is posed by the new concept of Web services: self-describing, self-contained, modular applications that can be mixed and matched with other Web services to create innovative and powerful distributed applications. Such systems will require the ability to conduct electronic transactions with other systems without any prior knowledge about their capabilities. Several major players, including Ariba, IBM, and Microsoft recently issued a specification for a Web Services Description Language (WSDL) which defines an XML grammar "for describing network services as collections of communication endpoints capable of exchanging messages." Same companies previously created Universal Description, Discovery and Integration (UDDI) standard, an ambitious core specification for business-to-business integration. Such systems will play an important role in the Web of the future: more on that in some of the forthcoming articles. And at the very end of this article, we shouldn't forget the newest DARPA initiative, DAML (DARPA Agent Markup Language), developed as an extension to XML and the RDF. This program formally began with a kickoff meeting in August 2000 in Boston. An integration contractor and sixteen technology development teams are working to realize the DAML vision. The latest release of the language (DAML+OIL) provides a rich set of constructs for creating ontologies and marking up the information so that it is machine readable and understandable.