Lists Home |
Date Index |
Chiusano Joseph wrote:
> XML Schema thus provides a form of semantics to the XML document, i.e.
> that such element content is to be treated as a decimal number.
> Yes - and in your opinion would the following XML snippet:
> provide a form of semantics to the data - i.e. that the data represents
> the estimated amount that an applicant requested, given that it occurs
> in the context of - for example - a loan application? This added context
> would therefore enhance the sematics provided by the
> "ApplicantEstimatedAmount" tag to enable it to be interpreted as "the
> amount that an applicant requested for a loan".
I am not saying that given sufficient _other_ information, that a program
cannot so interpret the snippet ... for example, application level
semantics. On the other hand, the XML Schema specification does not provide
> It doesn't state what units, for example, the number is intended to
> represent, e.g. 12.3 froggets, or 12.30 euros.
> Yes - but what if it did? For example:
Sure, specifications can define semantics for XML. The RDF/OWL semantics do
so for XML that conforms to the RDF/XML syntax. Writing a model theory is
one way to formally define a particular semantics. Other semantics might be
specified in the english language prose of a specification. That too is a
semantics, just not one that machines do a great job of properly
understanding. That why we still need human programmers to translate the
semantics of a written specification into a (machine) application. Much of
the idea behind creating machine processable ontologies etc. is that this
allows generic programs, e.g. inference engines, to do more, just as
computer languages greatly facilitate the porting of applications to
different platforms, and even, in the case of a more comprehensive semantics
such as is defined by the java virtual machine, enable applications to (very
frequently) operate across platforms without the need for human tweaking of
the application for a given platform/CPU.