RDF Metadata and Agent Architectures

Ora Lassila

Nokia Research Center (and W3C*)
3 Burlington Woods Drive, Suite #260
Burlington, MA 01803


This paper introduces RDF, the Resource Description Framework, as an emerging standard for metadata on the World Wide Web. RDF, if accepted by the web community, can provide simple knowledge representation facilities for the web. We argue that distributed applications of the future, possibly based on multiagent technologies, not only would greatly benefit from universally available facilities for knowledge representation, but actually require them to function productively.

* The author is currently working as a visiting scientist with the World Wide Web Consortium.


Metadata is "data about data" and can be thought of as (possibly formal) descriptions of some other data (which may or may not be available on-line). This paper discusses applications functioning in - or taking advantage of - the World Wide Web, so metadata specifically in this context is "data describing web resources."

The Resource Description Framework or RDF [Lassila & Swick 1997, Lassila 1997] is an emerging interoperable standard for metadata on the web, currently being worked on by the World Wide Web Consortium (W3C). Metadata can be seen as a possible solution to one of the major shortcomings of the World Wide Web as it exists today: It is really hard to automate any tasks which one has to perform on the web. So far, the web has mainly been built as a forum for human interaction; because most web documents are written for human consumption, the only available form of searching on the web - as an example - is to use simple string matching. Anyone who has used a web search service knows that typing in a few keywords and receiving a couple of thousand "hits" is not necessarily very useful. A lot of manual "weeding" of information has to happen after that; it may also happen that the keywords for which you are searching are not prominent in the relevant documents themselves.

A proposed solution for the search problem - and for the general issue of letting automated agents roam the web performing useful tasks - is to provide a mechanism which allows a more precise description of those resources that are available on the web. This, in turn, could elevate the status of the web from machine-readable to something we might call machine-understandable. As our proposed mechamism, RDF - the Resource Description Framework - is a foundation for processing metadata; it provides interoperability between applications that exchange machine-understandable information on the web. RDF emphasizes facilities to enable automated processing of web resources. RDF metadata can be used in a variety of application areas; for example: in resource discovery to provide better search engine capabilities; in cataloging for describing the content and content relationships available at a particular Web site, page, or digital library; by intelligent software agents to facilitate knowledge sharing and exchange; in content rating; in describing collections of pages that represent a single logical "document"; for describing intellectual property rights of Web pages, and in many others. RDF with digital signatures will be key to building the "Web of Trust" for electronic commerce, collaboration, and other applications.

At the core of the issue of making agent architectures (as well as distributed object architectures in general) work on the web is the ability to describe things for the purposes of interoperability. As a universal description language - sort of a simple knowledge representation system - RDF could be used for communicating interfaces of objects, capability models, and like.

Why Agents?

Agent technologies - especially multiagent technologies (where communities of agents interact with each other) - offer great promises of more capable, more flexible distributed applications. The ability to interact with other remote systems and to perform tasks autonomously without constant communication with the user are characteristics one would expect from systems and applications addressing - for example - the massive volume of information on the World Wide Web.

Certain basic mechanisms are a prerequisite for agent systems to work. Most importantly, agents need to be able to communicate with each other. Invocation mechanisms provide the basic capabilities for this requirement: agents can use current distributed object services (such as those built using RMI or IIOP) or new invocation protocols under development like HTTP-NG [HTTP-NG].

The emergence of wireless technologies places a requirement on agents to be able to function autonomously perhaps for long periods of time. Wireless communication by its nature requires communicating parties to be able to handle situations where continuous, uninterrupted communcation can not necessarily be relied on. Even in this area, basic infrastructure is finally emerging (for example, in the form of the Wireless Application Protocol [WAP]).

Why Knowledge Representation?

Mere transportation of bits is not sufficient communication for agents. Agents need to be able to communicate in a high-level language (not necessarily just one language), and they need to represent complex concepts. This is where we see RDF as an invaluable infrastructure.

RDF in itself will offer an extensible type system which allows one to build class hierarchies and - due to RDF's expressive capabilities - can be used to specify ontologies or term vocabularies. This places RDF as a language which agents can use for describing their capabilities and negotiating the terminologies used in communication. The road to agent architectures requires description mechanisms, and RDF could be used in conjunction with other agent languages (such as KQML [Labrou & Finin 1997]) to handle complex representational tasks. In fact, in many cases RDF could be used to substitute KIF [Genesereth 1995] as a more broadly understood - albeit less expressive - form of representation.

We see the future of distributed object applications to be built using various multiagent techniques. It is essential, however, that agents are supported by strong mechanisms for describing not only the agents themselves and their capabilities, but also other resources on the web. Resource discovery by agents can enable qualitatively more flexible applications than those in existence today, due to the fact that systems can be built to intelligently react to situations and environment not known at the time of system design.


The World Wide Web suffers from various problems, including the fact that it has been built for humans and not for automated systems to use. RDF is emerging as a way of elevating the level of machine-understandability on the World Wide Web. This, in turn, may lead to new opportunities for various automated systems - such as intelligent agents - to use the web as a vast information resource.

In the world of a rapidly changing web, distributed systems built to take advantage of it have to be built in a very flexible manner. Agent technologies may offer solutions to realize these systems, allowing applications to intelligently react to the creation of new information sources, changes in the topology of the web, and other possible opportunities and problems.


[Genesereth 1995] Michael R. Genesereth, 1995. Knowledge Interchange Format Specification. ANSI X3T2 working draft. Available on-line as http://logic.stanford.edu/kif/specification.html.

[HTTP-NG] --, 1997. Hypertext Transfer Protocol - Next Generation. Available on-line from http://www.w3.org/Protocols/HTTP-NG/.

[Labrou & Finin 1997] Yannis Labrou & Tim Finin, 1997. A Proposal for a new KQML Specification. Report TR-97-03, Computer Science and Electrical Engineering Department, University of Maryland Baltimore County. Available on-line as http://www.cs.umbc.edu/~jklabrou/publications/tr9703.ps.

[Lassila & Swick 1997] Ora Lassila & Ralph R. Swick, 1997. Resource Description Framework (RDF) Model and Syntax. Working Draft, World Wide Web Consortium. Available on-line as http://www.w3.org/TR/WD-rdf-syntax/.

[Lassila 1997] Ora Lassila, 1997. Introduction to RDF Metadata. W3C NOTE 1997-11-13, World Wide Web Consortium. Available on-line as http://www.w3.org/TR/NOTE-rdf-simple-intro.

[WAP] --, 1997. Wireless Application Protocol Architecture Specification. WAP Forum. Available on-line from http://www.xwap.com/.

Ora Lassila 1997-11-21