Information Technology Reference
In-Depth Information
To understand how an event-based API can work, we consider the
following sample document:
”1.0”?>
<doc>
<para>Hello, world!</para>
</doc>
An event-based interface will break the structure of this document down
into a series of linear events, such as those listed below.
start document
start element: doc
start element: para
characters: Hello, world!
end element: para
end element: doc
end document
An application handles these events just as it would handle events from a
graphical user interface; there is no need to cache the entire document in
memory or secondary storage.
In scientii c workl ow management, data-intensive applications are
typical. Should all data be calculated using some inner storage or i ltered
to reduce expense? We prefer the latter to build a i lter-like processor for
scientii c workl ow.
9.3.3
XML Data Stream View for Scientific Workflows
Specii cally, we can take an XML data stream view for scientii c work-
l ows from three different points: query tree, theoretic bounds, and event-
i ltering techniques.
9.3.3.1
Similar Tree Pattern Schema
It is well known that under the XML data format, the query can be
expressed as a tree structure [31]. Actually, a query tree is a part of an
XML schema for the XML documents or the XML streams. It has (or
partially has) functions like the schema for workl ow.
Because a workl ow schema is not a tree, to migrate the technique from
XML data streams we should segment the workl ow schema into different
layers in order to obtain schema trees and to apply our tree-based algo-
rithm to process it.
Figure 9.8
shows that there should be many strategies that can be
adopted for segmentation. Which one is a more suitable candidature
depends entirely on the applications and techniques to be applied. After
segmentation, we can use tools for processing the tree-structured queries
toward the scientii c workl ow such as automata (NFA/DFA).
Search WWH ::
Custom Search