Report prepared for the Experts Meeting Towards the Implementation of a Global Invasive Species
Information Network (GISIN), 6-8 April, 2004. Baltimore, Maryland, USA.
Page 37
8/30/2004
Web Services: Who, What,
Why, Where, and When?
Robert A. Morris, Professor of Computer
Science, University of Massachusetts/Boston,
Computer Science Department, 100 Morrissey
Blvd, Boston, MA 02466 USA.
Abstract
Database designers cannot know in
advance how their data will be used, or what
information a user or an application will need to
combine with query responses, much less where
that information will come from. The Web
Services architecture addresses these and other
questions that usually do not arise in centralized
databases and so supports more robust and
extensible systems than monolithic, centralized
data repositories. Using current open source or
proprietary tools, most of the work of deploying
Web Services comes down to reusing the
interfaces that already exist between databases
and Web browsers, and designing controlled
vocabularies to represent the content. Since a
controlled vocabulary is desirable in virtually any
data sharing architecture, the technical appeal of
Web Services is very high. In some
environments (e.g. Microsoft.net), the technical
burden on the organization deploying the service
is very low.
This is a non-technical overview of the
answers to the questions of the title as applied
to the Web Service architecture, along with how
the architecture itself dynamically addresses
those same questions for distributed data
sources. The huge existing base of Web Service
tools and standards offers an appealing way to
position GISIN to concentrate more on content
and end-user applications than on infrastructure.
Its natural fit to a model of open data adds
further appeal. This paper briefly outlines what
would be necessary for GISIN to be based on
the Web Services technologies, and what parts
of the GISIN enterprise are not in this realm and
would have to be accomplished no matter what
distributed database environment is
implemented.
Introduction
The most important thing for a data provider
to understand is that it is impossible to know
or predict who is going to use the data and
how they are going to use it. The second
most important concept to be grasped is the
fact that software does not know anything
about anything. All software is ignorant, but
both informed and uninformed users are
nevertheless often well served by such
software. A third issue is that it is very
difficult to take something that was originally
designed for human consumption or
interpretation, and put a machine interface
on it. It is however very easy to do this in
the reverse.
… it is impossible to know or predict
who is going to use the data and how
they are going to use it.
Finally note that from the informatics point
of view, there isn’t any difference between
data and metadata. They are all treated in
the same way.
The aim of this report is not to provide a
technical discussion of Web Services.
Instead it provides an idea of what the
GISIN participants might need to be
concerned about including, what they will
get out of using Web Services, and how
hard it may be to ask data managers to use
Web Services.
What is a Web Service?
A web service is any facility on the Web that
can answer a question or provide some sort
Point of View
•
A data provider can not know how its data will be
used
•
Software does not know anything about biology.
•
Software does not know anything
about
anything.
–
Design for ignorant software; informed (or
uninformed) users will be well served.
–
When Morris uses pronouns (you, me, I, they)
he almost always means software, not
people.
–
There is no difference between data and
metadata. Both are first of all for software
before humans.
 |
 |
 |