[X-logic-www] Re: [X-logic-arch] On Portal Specifications (X-Logic)

[Neil N. <n_n...@pa...>]

Roger,

In not being familiar with XML I took the 5 minute introductory
tutorial at http://java.sun.com/xml/tutorial_intro.html from which
I now have just enough knowledge to be dangerous.  My
impression of XML is that it is a data description language
with the advantages of: being oriented to web pages,
having a good and growing market share, being likely easily
used with Java.

A missing desirable component for XML available to most
database languages is a data access utility, which may be
one of your interests in developing X-logic, in that logic
operations on data via a data description language would
give the  usual data access utility the would be typified by
SQL.

Though it would be inefficient with respect to say, MySql,
to store the more commonly uniform data in XML there
does not seem to be any necessary objection.  And
perhaps there is interest in using the XML data description
language for, say, SQL data and then have an SQL
like language, X-logic, that could be arbitrarily applied to
XML descriptions.

Since XML data is intended to be web-page text it will
not typically be sufficiently uniform to apply the more detailed
SQL commands such as equality and order relations to
that data.  However, equality and order relations could be
applied to the data description elements in XML that would
then order and group classes of data according to that XML
description.  And then as certain data elements are noted as
useful for the detail relations, that increases, in a sense, the
cardinality of the model.  I.e., the cardinality of the model
may be described as the number of distinct elements that it
has, and such that the usual cardinality of uniform data is
high because of the usual easy separation of each element.
Whereas if we could only address the data description, the
cardinality would be low as represented by the smallest
model of that description, which would be the nature of
an XML description with arbitrary text elements in a web
page.  And then there would be intermediate cardinalities/
models where some portion of the data was uniform that
would add properties for the logic/SQL to grab onto.

Just beyond the immediate logic application there is some
potential to apply standard techniques to text data; e.g.,
common search/index methods on key words and
phrases could be used to obtain more properties.  This
area could be extended to cover natural language
processing that would in a model sense maximize the
commonly available properties.  I.e., there would be a
natural model cardinality break point between commonly
and specifically used languages (syntaxes). And then
specific languages such as in mathematics or other, say,
computable languages could also be addressed.  But
in moving beyond the immediate logic and order
relations the number of potential relations increases
faster than any feasible general method that then
requires specific coding to capture those relations
useful at the moment (the application specific functions).

This is just my interm global framework for what we
are trying to do with X-logic.  Does this seem to conform
to the way you are thinking?

Regards,

Neil Nelson