Issue Request: The unqualified term "Semantics" should be avoided

The recent discussion on www-tag has demonstrated that the term "semantics", 
used in an unqualifed way, is unhelpful to intelligent, precise and informed 
discussion of architectural issues. The diametrically opposed views of 
whether generic XML or XHTML lack or are rich in "semantics" demonstrates the 
difficulties of using the unqualified term.

I would like to ask TAG to consider the issue of whether the unqualified use 
of the term "semantics" lacks clarity and is harmful to good communication 
and W3C specification development. Since W3C is placing significant emphasis 
on a "Semantic Web" clarity about which types of semantics are considered of 
value seems important, not least in that strategic context.

I would also ask that TAG considers whether W3C Working Groups should be 
encouraged or instructed to avoid the use of the unqualified term "semantics" 
in public specifications. This could aid clarity of specifications, something 
that is notably lacking in several.

I would further request that TAG gives consideration to formulating a less 
ambiguous terminology to describe the forms of semantics which might enter 
into scope for W3C technologies and encourage W3C specifications to adopt 
relevant terms.

The following suggested structure is tentative and open to change. Often it 
is easier to gain clarity when seeing a proposal that is clearly wrong or 
differs from one's own perceptions in how it classifies terms.

It seems to me that the broad term "semantics" might be considered as 
covering the following general areas:
1. Element Type Name Semantics - hints to meaning potentially, but not 
necessarily, provided in the start tag
2. Structural (contextual?) semantics - found in or guided by schemas
3. Namespace semantics - the additional clarity about the meaning of, for 
example, an element provided by a namespace e.g. compare <stylesheet> vs <
xsl:stylesheet> (forgive the shorthand of using a namespace prefix)
4. Metadata semantics - semantic information provided by metadata, RDF etc
5. Presentation (display?) semantics - as found in HTML and XHTML
6. Styling Semantics - provide, for example, by CSS
7. Visual semantics - e.g. as found in bitmap graphics. It is untrue to say 
that these images lack semantics. The seeming absence of semantics is a 
reflection of the primitive nature of current software and its inability to 
recognise the semantics.

In addition, I suggest that imprecise terms such as "semantic markup" be 
avoided. For example, I don't reliably understand the term as used by Tim: 
"the TAG should consider a general statement of the value of semantic 

Is there semantics-free markup? If so, what is it?

Is "semantics-free" markup generic XML which takes no practical advantage of 
the potential to use Element Type Name Semantics? If so should XML 1.0 
encourage the sharing of generic markup on the Web?

If TAG intends to comment on "semantic markup" which type(s) of markup are 

Adding additional information about what type(s) of semantics are being 
referred to would aid clarity.

Andrew Watt

Received on Tuesday, 20 August 2002 03:57:44 UTC