Initial version: 1999-11-12, Dan Brickley
danbri@w3.org
Revised: March 2001
Status:
This is a work in progress, and an early release of the document
for feedback from the RDF Interest Group. It is intended as an
informal discussion document, and is not a formal
publication of any working group, or of the W3C itself. See the Semantic Web Activity pages for
information about current W3C work in this area.
Although the HTML is valid and the RDF/XML parses successfully with SiRPAC, W3C's Java RDF parser, readers are cautioned that the Javascript demonstration referenced here will only work on some platforms. Work is in progress on a server-based version of the demo that will allow for universal access from all HTML browsers.
General comments on this work-in-progress should be sent to the RDF Interest Group; bug reports should be sent to the author. More work is needed on XSL and SVG representations - assistance is welcomed...
The original proposal of the WWW from 1989 included a figure showing how information about a Web of relationships amongst named objects could unify a number of information management tasks.
This document describes a re-expression of this figure using W3C's Resource Description Framework datamodel in XML syntax.
The original figure, included here as a GIF image, depicts a Web of objects, including people, organizations, technologies, documents and topics. Typed links (such as 'wrote', 'unifies', 'includes' are used to represent knowledge about their interelationships).
(CERN figure
from WWW proposal)
The aim of this current document is to begin an exploration of how far the WWW has progressed since 1989, and whether recent developments (eg. XML, RDF, URI schemes, XSL transformations, SVG graphics, XML Linking) can be combined to address the issues set out in the original WWW proposal.
The XML/RDF document proposal.rdf that accompanies this document contains embedded comments intersperced with XML-encoded RDF statements. In translating the WWW proposal figure into RDF, various assumptions and ad-hoc decisions were made. These raise topics for consideration in the RDF community...
Disclaimer: Some familiarity with the RDF data model is currently presumed here; future versions of this document may be more accessible to a broader readership.
The initial exercise was as follows: create an RDF representation of the figure from the original WWW proposal
When assigning nodes from this diagram into RDF 'types' or classes, and associating the link-types with RDF 'properties' (ie. relationship types), some arbitrary decisions were made. The original link shown as 'wrote' was mapped to an inversely-directed link 'creator' defined by the Dublin Core Metadata Initiative. The links sketched as '...(for)example...' in the original figure have been interpreted as mapping to RDF's notion of 'type'. The link called 'describes' has been mapped (a little innacurately) to the Dublin Core's notion of 'subject'. The entities listed (CERN, VAX/NOTES etc) have been assigned arbitrary (and fictional) URIs for the purpose of this identifier. All other constructs not easily mapped to current RDF vocabularies have been assigned to a single and equally ficitious RDF vocabulary, whose schema has [@@TODO] not yet been written. Some portions of the diagram have for simplicity been ommitted (eg. the CERN hierarchy).
Having re-represented this data in an RDF format, what can we do with it that we couldn't before? The RDF pages list a number of query and logic oriented applications that suggest approaches to WWW knowledge management unavailable in 1989. For example, we can show a simple Javascript-based RDF Query demonstrator that queries this RDF database. (note that this is an in-progress work and currently functions in only a subset of Javascript/ECMAScript browsers). It is interesting to revisit the initial aims of the WWW, and to compare these with the facilities specified in recent W3C Recommendations such as RDF and XML.
Other documents accompany this overview: