W3C

RDF in XHTML

Task Force Document 27 May 2003

This version:
TBD
$Revision: 1.36 $ on $Date: 2003/08/18 13:22:07 $ GMT by $Author: reagle $
Latest version:
TBD
Previous version:
NA
Editors
Joseph Reagle <reagle@w3.org>
Dominique Hazaël-Massieux <domdw3.org>

Abstract

This document attempts to provide work plan for advancing the state of including RDF metadata in an XHTML document. This plan presently includes requirements, issues, and links to resources.

Status of this document

This document is a draft of the RDF in XHTML Task Force (charter) between the Semantic Web CG and HTML WG. (This issue is also W3C TAG issue RDFinXHTML-35, though it is not presently a high priority in that forum.)

This document does not necessarily represent consensus; it may include provocations, annotations and contrary positions (or alternative wordings) in order to elicit review and discussion. Positions which are potentially in conflict are specified as a list of lettered points. For example:

  1. Requirement
    1. Position
    2. Alternative/Contrary Position
    3. The currently dominant/favored position

      Evidence or a scenario that the requirement is compelling.

      Existing solution which that already satisfies a requirement.

    4. A discarded position

    Comments from the author, or points requiring further discussion.

Please send comments to this document to public-rdf-in-xhtml-tf@w3.org (publicly archived).

Table of Contents

  1. ...

Introduction and Goals

There is a long standing requirement to embed meta-data in an HTML document. One would think that this requirement could be satisfied by combining XHTML with RDF/XML using XML Namespaces and XML Schema, but this is not so. Instead, there are many nuanced technical issues and a series of problem statements, such as Tim Berners-Lee's RDF in XHTML (April 2002) and Dan Connolly's more recent RDF in XHTML (March 2003), and related proposals and analysis, such as Sean Palmer's exhaustive RDF in HTML (May 2002). Previous discussion failed to converge upon a solution. Furthermore, there hasn't been a coordinated plan for addressing this problem that encompasses the many constituencies. Consequently, the goals of this document are to:

  1. Identify the requirements and constraints for embedding RDF in XHTML.
  2. To document a solution for satisfying those requirements.
  3. To document the commitments and milestones necessary of each constituency to deliver the solution.

Scenarios and Constraints

The following scenarios and constraints have been presented as specific sources of requirements (beyond the more general references of the EmbeddingRDFinHTML Wikki):

FOAF
FOAF in XHTML Files, Dan Brickley
Trackback
Scenario: Trackbacks, Joseph Reagle
Dublin Core
Scenario: Dublin Core Dave Beckett

Existing Solutions

The following applications have been presented as already satisfying requirements in this document; and consequently have a deprecating effect on those requirements:

Dublin Core
Scenario: Dublin Core Dave Beckett. Dublin Core already mechanisms for encoding flat characteristics of the document in HTML meta elements.

Requirements

One can attempt to solve a problem by (1) throwing solutions at a problem and seeing if one "sticks" and (2) working from agreed to requirements. The proposed solutions so far have not yet stuck, but I believe this is because of the lack of a coordinated plan. I hope to roughly scope the problem with requirements, as augmented with evidence and scenarios, and then move to various solutions in the context of a specific plan. But first a note on scope, associating metadata with an XHTML instance via a link is out-of-scope for this document.

  1. Embedding, the solution must specify how
    1. To embed RDF metadata within an XHTML document
      1. Expressivity
        1. The solution MUST only support simple (flat) assertions — a single resource with one layer of properties. [Trackback] [Dublin Core] [Dublin Core]
        2. The solution MUST support arbitrary (rich) assertions. [FOAF]
      2. Format
        1. The RDF MUST NOT have to be reformatted from RDF/XML. [FOAF]
        2. The RDF MUST be reformatted in to a new syntax (such as in meta name attribute values)
      3. Scope
        1. The RDF MUST not inherits any semantic context from its container document.
          1. If it is metadata about that document, it should state rdf:about=""[FOAF]
          2. If it is metadata about a fragment, it should state rdf:about="#frag_id" [Trackback]
        2. The RDF MUST be able to inherit semantic context from its containing document.
    2. To embed arbitrary XML in XHTML
    3. To compose arbitrary XML in XML
  2. Authoring, the solution MUST enable
    1. XHTML authors, if changes are required of it, to by
      1. easily represent characters (such as those captures by MathML expressions and XHTML character entities)
      2. define arbitrary "entities" (as is presently permitted in XML with DTDs)

      XML nor XML Schema can satisfy the XHTML requirements without DTDs and it is posited that a new version of these specifications would be required.

    2. RDF authors MUST rely upon tools or copy/paste an arbitrary serialization into an XHTML document (hand authoring the RDF serialization is NOT a requirement)
  3. Validation
    1. The XHTML document
      1. (unknown elements are not ignored) all elements in an instance MUST be strictly validated against its XHTML definition.
      2. (unknown elements are ignored) all XHTML elements MUST strictly validate against its XHTML definition but other elements MAY be ignored.[Trackback] [Dublin Core]
    2. The RDF metadata
      1. MUST be valid XML.
      2. MAY NOT be valid XML but MUST be valid RDF/XML.
      3. MUST be well-formed XML but need not be valid XML (RDF validation is left to the discretion of the RDF application.)

    XML Schema can satisfy XML strict and lax validation requirements.

  4. Compatibility, the solution
    1. MUST not be rendered by existing browsers (the solution MUST work with all existing browsers)
    2. SHOULD NOT be rendered by existing browsers

Deliverables

The dominant set of requirements would be facilitated by:

  1. Possible ACTION XML Core or XML Schema WG: develop a specification (of XML or XML Schema) that specifies a feature akin to character entities without DTDs. This would permit the HTML WG to move towards relying upon alternative schema languages (e.g., XML Schema and RelaxNG) that would then be more permissive with respect to including foreign XML (including embedded RDF 1.0 serializations).

    The following history was investigated to determine if any existing commitments are pending (most links are to Member only resources):

    Consequently, it appears the Schema WG has a pending action to satisfy the HTML WG, and the Core WG has a pending requirement for XHTML 2.0. However, I expect neither WG recognizes such a commitment and recommend it be re-raised. Absent the explicit designation of resources to solve this issue, I simply expect many formats (e.g., HTML, MathML) will continue to rely upon DTDs.

  2. Possible ACTION RDF Core WG: develp a specification of a 4.4 Normative specification of XML grammar for RDF that can be validated by XML Schema, as requested by the XML Schema WG. This would permit embedded RDF to be validated as part of an XHTML document.

    The RDF Core WG has declined/postponed this request. When would this request likely be raised again? The Semantic Web Activity Statement states that, "An Activity proposal for Phase 2 of the Semantic Web reflecting this and other work ready for standardization is expected to be available to the Membership in Q4 2003." Unfortunately, this would be rather late as the HTML WG is hoping to last call their draft of XHTML 2.0 in the summer of 2003.

  3. Possible RDF Core / HTML WG: develop a specification of a new or "not-so-strictly conforming" XHTML document that permits RDF 1.0 serialization within an XHTML document.

    Joseph Reagle has deployed an temporary validator service using Masayasu Ishikawa's experimental schema for XHTML+lax-foreign-namespaces. However, this solution will not be able to use character entities absent a deliverable from (1) above.

Milestones

2003 Jul
Opened lines of communication and plans for progress.

Resources

EmbeddingRDFinHTML, Wikki. May 2003.

[RDFinXHTML-35] Syntax and semantics for embedding RDF in XHTML. Dan Connolly. March 2003.

Embedding RDF [and other XML vocabularies] within an XHTML Document. Ralph Swick. March 2003

"XML Schema for RDF", Rick Jelliffe, February 2000; and derivative XML Schema for RDF and RDFS, Dan Connolly, July 2000.

"RDF in HTML", Tim Berners-Lee, April 2002.

"RDF in HTML: Approaches", Sean Palmer, June 2002.

"Metadata for Grandma", Micah Dubinko, September 2002.

A strawman Unstriped syntax for RDF in XML, Tim Berners-Lee, May 1999.

Appendix B to 1999 RDF spec, February 1999.

Metadata for Grandma, Micah Dubinko, September 2002.

XML in HTML Meeting Report , Dan Connolly, Lauren Wood, 11 May 1998.