Review of Concepts Doc

Here are my comments regarding the Concepts Doc. Issues
that I consider critical, to be addressed before actual last call
are prefixed with asterisks. Other comments may be considered
non-critical, but the editors are certainly welcome to address them
before last call, time permitting.

1. Introduction

It may be better to say "RDF Base" rather than "RDF Core" in
order to (a) sync with the terminology/organization of the
XML specs and (b) avoid confusion with "RDF Core WG".

1.1 Structure of this Document

The text

"RDF's abstract syntax is a graph, which can be serialized using XML (but which is quite distinct from XML's tree-based infoset
[XML-INFOSET]). The abstract syntax captures the fundamental structure of RDF, independently of any concrete syntax used for
serialization."

should probably be moved to just before section 1.1 as it has nothing to do with
the structure of the document though is certainly introductory material.

2. Motivations and Goals
2.1 Motivation

*** The statement "RDF provides a world-wide lingua franca for these processes."
*** is incorrect. RDF is not itself a language for system interaction. It is
*** a meta-language for defining languages for system interaction. Just as XML is
*** a meta-language for defining document models and is not itself a document model.
*** This should be rephrased as something like "RDF provides a common foundation for
*** the reliable interchange of information between such processes."

2.2 Design Goals

Given the ambiguity of the word "syntax", it might be better if
the bullet "XML-based syntax" would rather be "XML-based serialization
syntax" (as it is in section 3).

2.2.1 A Simple Data Model
2.2.2 Formal Semantics and Inference
2.2.3 Extensible URI-based Vocabulary
2.2.4 XML-based Syntax
2.2.5 Use XML Schema Datatypes
2.2.6 Anyone Can Make Simple Assertions About Anything
2.2.7 Arbitrary Expression of Simple Facts
2.2.8 A Basis for Binding Agreements
3. RDF Concepts
3.1 Graph Data Model
3.2 URI-based Vocabulary and Node Identification
3.3 Datatypes (Normative)

*** No mention is made in this section of datatypes defined in
*** frameworks other than XML Schema. It needs to be clearly
*** stated that while RDF Datatyping adopts foundational concepts
*** from XML Schema Datatyping, and thus XML Schema datatypes are
*** (generally) usable with RDF, RDF Datatyping is compatable with
*** any datatype conforming to the defined characteristics of
*** rdfs:Datatype and need not be defined (or even definable) by
*** XML Schema. Since this seems to be the most substantive normative
*** definition of RDF Datatyping, this should be made clear here.

3.4 Literals

The first sentence

"Literals are used to identify values such as numbers and dates by means
of a lexical representation."

seems only to apply to typed literals, given the language of 'values'
and 'lexical representation'. Perhaps a more generic statement is in
order, such as

"Literals are used to identify resources such as strings, numbers,
dates and XML encoded content which in RDF have a non-URIRef textual
representation."

-

Additional clarification of "plain literals ... are self denoting"
would be useful, to bring home the fact that the plain literal
"25" cannot be interpreted to mean twenty-five.

3.5 Representation of Simple Facts

The text

"Some simple facts indicate a relationship between two objects.
Such a fact may be represented as an RDF triple in which the predicate names
the relationship, and the subject and object denote the two objects."

is confusing in its use of 'object' with two different senses. Better to
use 'resource' for the first sense. E.g.

"Some simple facts indicate a relationship between two resources. Such a fact
may be represented as an RDF triple in which the predicate
names the relationship, and the subject and object denote the two resources."

Since everything in RDF is a resource, including literals, this is clearer.

-

The first real example of the abstract graph syntax seems to be a
rather complex one, bringing in relational tables and class typing
immediately into the mix.

It might be good to first have a more basic example earlier on, without
blank nodes, rdf:type, etc.

3.6 Entailment
3.7 RDF Core URI Vocabulary and Namespaces

The statement "RDF uses URIs to identify resources and properties." should probably be
shortened to just "RDF uses URIs to identify properties." since (a) this section is
about vocabulary (properties) and resources may also be identified by literals.

4. Meaning of RDF (Normative)
4.1 Asserted and Non-asserted Forms
4.2 Social Meaning

Change "...as an assertion in any other format." to
"... as an assertion in any other form."

4.3 Authoritative Definition of Terms
4.4 Interaction Between Social and Formal Meaning
4.5 Example (Informative)
5. XML Content within an RDF Graph (Normative)

I am uncomfortable with the <rdf-wrapper> convention, but as
it seems to work, so be it. Let's see what the community has
to say...

*** However, it seems odd that we would not be fully compatable
*** with XML 1.1 if at all possible. I.e., should we not say that
*** all XMLLiteral values should be normalized?

6. Abstract Syntax (Normative)
6.1 RDF Triples
6.2 RDF Graph
6.3 Graph Equality
6.4 RDF URI References
6.5 RDF Literals
6.5.1 Literal Equality
6.5.2 The Value Corresponding to a Typed Literal

In the Note:, you could make it clear why it is more useful
to compare values rather than lexical forms -- that a given
value may have more than one lexical representation and therefore
a comparison of forms may return F whereas a comparison of
values may return T.

6.6 Blank Nodes
7. Fragment Identifiers
8. Acknowledgments

*** Sergey is mentioned in the first paragraph but not the second. I'm
*** presuming this was an unintentional omission.

9. References
9.1 Normative References
9.2 Informational References

[Patrick Stickler, Nokia/Finland, (+358 40) 801 9690, patrick.stickler@nokia.com]

Received on Monday, 30 December 2002 08:17:26 UTC