untitled

200 9 The Transformation Process

ment by simply changing one line of the document. In a similar way, one can

specify the style of an XML document using astylesheetas follows:

<?xml version="1.0"?>

<?xsl-stylesheet type="text/xml" href="transform.xsl"?>

In the example above, the XML document would be transformed by the

stylesheet file named “transform.xsl.” The stylesheet is a separate file just

as in LaTeX. There are many stylesheet languages for XML. Because of this

history, XML transformation programs are often called “stylesheets” even

when the transformation has nothing to do with presentation style.

The three approaches to transformation are covered in the next two chap-

ters. Event-based parsing and tree-based processing are covered using the

Perl programming language in subsections 10.2.2 and 10.2.5. Rule-based

transformation is covered by using the XML Transformation Language in

chapter 11.

Summary

A transformation step is performed using one of three main approaches:

1. Event-based parsing


2. Tree-based processing


3. Rule-based transformation

9.5 Automating Transformations

So far we have been assuming that transformations between ontologies will

be specified by people, usually domain experts. It is natural to suppose that

when two ontologies refer to the same concepts, it ought to be possible to

transform from one ontology to the other using some automated process

without the need for human effort. The problem of automating the process

of finding semantic correspondences between different ontologies has been

studied for many years. Most of the work has been for relational database

schemas, but there has also been some recent work on this problem for XML

DTDs and even for the more sophisticated ontologies of the Semantic Web.