RDF Primer Comments

These comments are on the 19 March 2002 working draft of the RDF Primer
http://www.w3.org/TR/2002/WD-rdf-primer-20020319/

They are in addition to the previous comments
http://lists.w3.org/Archives/Public/www-rdf-comments/2002JanMar/0238.html

(Although I think a section on applications is more than marketing, it 
will help understand the use and value of RDF so I would keep that section).

and

http://lists.w3.org/Archives/Public/www-rdf-comments/2002JanMar/0240.html


The primer is a very good document and very helpful. I have a few 
suggestions and comments:

1. I would suggest adding a section between the use of XML (2.2) and the 
RDF model (2.3) on "Establishing Meaning: Ontologies etc". The points 
would be:
- an ontology is used to establish a common framework for understanding 
of terms and their meaning, both denotations and connotations (is that 
right?)
- Maybe explain the relationship of DAML to RDF, or the role of DAML.

2.In the URI section mention the relationship between a URI for a 
resource and a possible corresponding web document. This is a confusing 
issue discussed on the list that might deserve a sentence or two.

3. Explain the relationship of XML Schema to RDF, and the relationship 
between XML Schema and RDF Schema.

At the end of section 3 (XML syntax for RDF) explain how the XML syntax 
may be defined using XML Schema, used to define XML content types and 
legal XML element and attribute structures. Introduce RDF schema by 
indicating it is more focused on defining classes and appropriate RDF 
relationships.

XML Schema is about legal XML representations and data types, RDF schema 
is more about constraints on object relationships.

4. Use of triples is helpful, sometimes easier to understand than graphs.
Maybe mention N3 when introducing triples, and refer to it in tools 
section. N3 does a lot to make RDF accessible, I'd almost argue that n3 
should be the primary "view" of it.

5. I admit I have work to do reading the various RDF documents, but this 
introduction might need a little more explanation of why striping 
matters. Perhaps move the reference to STRIPEDRDF earlier to where 
striped is first mentioned. Why does breaking striping matter?

6. Why do abbreviations matter with XML serialization of RDF? If I want 
a short form, won't I be using n3 anyway (not for interoperability)? I 
think the section on abbreviation needs some motivation.

7. Property attributes could use an example (like in the RDF spec I believe)

8. what does parseType="Resource" mean in the example in section 3?

9. I believe tools deserves its own section at the end, as well as a 
place in the table of contents. N3 should be included.

10. Why is an RDF resource either a string literal, URI, or blank? 
Couldn't any XML schema type be used (e.g. dateTime)?

11. Model theory 6.1 seems out of place in this primer. It seems there 
needs to be more rationale and benefit of model theory and an example to 
  motivate the paragraph.

The Minivan example is good.

Some detailed comments:

12. Section 3 - should we say QName instead of local name for the name 
used for properties and objects?

13. Section 3 paragraph 2 sentence might be better changed from

The Node at the start of the sequence is always a subject node and turns 
into a containing element called an rdf:Description that is written at 
the top level of RDF/XML, after the XML document element (in this case 
rdf:RDF).

to

The sequence starts with a Description Node corresponding to the 
starting subject. This rdf:Description element is contained in an outer 
rdf:RDF element.

14 In section 5.3, the following fragment seems to be missing some text:

Without such protection, the company's networks will soon collapse under 
the load or its clients will consider themselves willfully "spammed" and 
withdraw their custom.

I'm not sure what "withdraw their custom" means.

15.Could some examples for the following statement about "most modern 
cases" be added?

The technology concerned is "routing" and, in the most modern cases, 
relies on RDF.

16. the following does not make sense:

Judgments about distributing material can be made in a context values 
(the standard predicate systems like Dublin Core) and a vast number of 
alternatives,

Does "in a context values" mean "according to a meaning framework 
specified in RDF" ?

17. At the end of 5.3, what is to prevent a spammer from creating RDF 
Spam? Perhaps a statement about the use of XML Digital Signatures in 
conjunction with RDF to achieve this goal would be useful.

"Combining XML digital signatures with RDF descriptions to ensure that 
you only receive desired information from appropriate sources should 
lead toward the elimination of spam."

18. Is there an issue of the quantity of meta data which will make it 
hard to use?

I hope this is helpful.

< Frederick

Frederick Hirsch
hirsch@fjhirsch.com

Received on Friday, 26 April 2002 15:48:16 UTC