100 likes | 204 Views
“I Was There…” – Memoirs of an RDF Working Group Member or Observations about the RDF Design Rationale. Ora Lassila Research Fellow, Nokia Research Center December 2000. RDF Timeline. 1997 Spring Pre-WG work (e.g., PICS-NG), “authors’ meeting” @ MIT
E N D
“I Was There…” – Memoirs of anRDF Working Group MemberorObservations about theRDF Design Rationale Ora Lassila Research Fellow, Nokia Research Center December 2000
RDF Timeline 1997 Spring Pre-WG work (e.g., PICS-NG), “authors’ meeting” @ MIT 1997 Summer M+S Working Group chartered, M+S first draft (for group review) 1997 October M+S first public draft 1997 November RDFS Working Group chartered 1998 April RDFS first draft (for group review) 1998 October M+S goes to “last call” 1999 January M+S goes to proposed recommendation 1999 February M+S goes to recommendation! 1999 March RDFS goes to proposed recommendation 1999 August RDF Interest Group formed 1999 October The “Cambridge Communiqué” published 2000 March RDFS goes to candidate recommendation
Motivation • library metadata (Dublin Core) • content rating (PICS) • site maps • some other applications…
Model • Ora (from PICS-NG): frame-like model • Guha: understanding which statements have been asserted, and which ones have not • WG charter included mandatory PICS support • certain features, which cannot really be implemented in the model itself, crept in (“aboutEachPrefix”) • Acceptance & deployment was very important • “just simple enough” for the WWW community at large to accept and deploy • “not too offensive” for the KR community so it could be used as a starting point for something better • main challenge: managing expectations
Syntax • Naming (e.g., “Pumpkin”) • S-expressions vs. XML • in some sense, the choice of XML was an unfortunate one, because it leads to a lot of confusion • Namespaces were deemed necessary, and consequently an XML NS spec which supports RDF needs was “rammed through” at W3C • issues with the namespace of attributes like “about” • Interpretation of literals • XML Schema was supposed to provide “primitive” datatypes
Details, Details, Details, … • RDF is supported by a number of other standards • XML • URI • HTTP (caching semantics) • … • It is important to understand that RDF takes care of a lot of “dirty details” which we now no longer have to worry about
Type System & Ontology - RDF Schema • Basic definition of “Class” • defined as a prototype rather than a classification • Metaclass issues proved to be hard • ANSI X3J13 as an inspiration, but simplified • class Class and “class Metaclass” are the same thing • “DisjointWith” and cardinalities: discussed but eventually rejected • Domain & range proved to be hard (for the WG) • “subPropertyOf” vs. “subClassOf”
Property type type type domain Class type type p c type y x “Mysteries” of Domain & Range
Other Issues • “Dueling press releases” • Netscape’s love for RDF vs. Microsoft’s marketing message • a lot of the RDF M+S work happened at the height of the so-called “browser wars” • WG member troubles • skill/experience vs. technical complexity mismatch • RDFS vs. XML Schema • cf. the “Cambridge Communiqué” • RDFS still not a recommendation…
Questions? • mailto:ora.lassila@nokia.com • mailto:daml@lassila.org yawn… Lauren Lassila (age 3 months) finds the RDF Design Rationale a perfect bedtime story.