W3C home > Mailing lists > Public > xmlschema-dev@w3.org > November 2011

Re: "XML Schema specifies just syntax" versus "XML Schema specifies semantics"

From: Steven Ericsson-Zenith <steven@semeiosis.org>
Date: Wed, 16 Nov 2011 22:27:01 -0800
Message-Id: <66683DBB-86C8-4CEF-8615-3A1F78389622@semeiosis.org>
To: xmlschema-dev@w3.org

It seems to me that the "Semantic Web" and much of the associated effort fails because of this lack of formal rigor.  Appealing to "common use" will only lead to heartache - application programs that either do not work or whose behavior surprises us.

On the other hand, xml-schema plays an important (formal) role in specifying the valid structure of XML documents. It enables the formal specification of valid XML structures. xml-schema does not specify the validity of operations upon these structures. 

XSLT can specify the formal transformation of these structures. XSLT is not perfect in this regard, but it can be used to formally specify application semantics. Without such semantic rules, rules of transformation, xml-schema alone does not specify a formal grammar. 

In Logic, at least, the term "meaning" can and often does have a rigorous definition. In the case that I am most familiar with it refers to the behavior produced by the processing of a formal grammar. The question of what does that processing, human or machine, is one of apprehension. If a human or machine cannot consistently "read" (apprehend) the document and thus apply these rules to a well formed document then that document either has no meaning or its meaning is unpredictable. I am allowing that you may allow ambiguity such that the apprehender may choose to behave as it likes but you can never reliably expect an apprehender to behave in a predictable way unless you remove this ambiguity.

Now your model of logic may vary but it will at least follow this kind of rigor.

You can define the term "ontology" to be "hierarchy of terms in a controlled vocabulary" but this is not a phrase that makes much sense to me. I know that in the past decade the "Semantic Web" and others in IT have been using this term but this use is completely unrelated to the historical use of this term. Some would say that this usage is "incorrect" but I will be happy if you just gave a rigorous definition of it that does not overload, or is at least consistent with, the concepts associated with the other terms that you use. As is, it isn't, it's a redundant concept, unnecessary. 

With respect,
Steven


On Nov 16, 2011, at 8:46 PM, Liam R E Quin wrote:

> On Wed, 2011-11-09 at 11:21 -0800, Steven Ericsson-Zenith wrote:
>> In Logic "semantics" refers to the rules of valid syntax and its
>> transformation. I can't make any sense of the definition given here.
> 
> I think Roger is using a mixture of the Semantic Web and linguistic and
> common-usage definitions, rather than that of logic and predicate
> calculus.
> 
> In any case I don't see it as helpful; it's clear that people can and do
> use XSD (W3C XML Schema) and other XML schema languages, to define
> ontologies, that is, hierarchies of terms in a controlled vocabulary;
> they can give them human- and machine-readable "meaning" e.g. using
> annotations and/or RDF.
> 
> An XML Schema defines a formal grammar, but also defines ways for a
> Schema processor to associate type and other annotations to XML markup.
> 
> If RDF can carry meaning, and if you can use a Schema to associate RDF
> with XML elements, it seems reasonable to imagine that you can use a
> Schema to impart meaning to an XML document.
> 
> Whether these antics are called semantics is another question.
> 
> Liam
> 
> -- 
> Liam Quin - XML Activity Lead, W3C, http://www.w3.org/People/Quin/
> Pictures from old books: http://fromoldbooks.org/
> 
> 
Received on Thursday, 17 November 2011 06:27:32 UTC

This archive was generated by hypermail 2.3.1 : Wednesday, 7 January 2015 14:56:19 UTC