Translator lets computers "understand" experiments

[David Farber <dave () farber net>]

Begin forwarded message:

From: Randall <rvh40 () insightbb com>
Date: June 7, 2006 9:48:10 AM EDT
To: Dave Farber <farber () cis upenn edu>, Dewayne Hendricks  
<dewayne () warpspeed com>
Subject: Translator lets computers "understand" experiments


Wednesday, June 07, 2006 at 9:46 AM EDT
 A framework for translating the write-ups of experiments into a format
that can be processed by computers has been developed by academics.

The new tool could revolutionise the way scientific papers are written
and help scientists make creative leaps, researchers say.

Computers already help scientists by performing complex calculations and
making information easier to analyse. But they are less suited to
analysing and comparing experimental results.

"Computers are not very good with natural language, they need to have
things as formalised as possible," says Ross King, a researcher at
Aberystwyth University in Wales, who developed the framework with
colleague Larisa Soldatova.

Called EXPO, it can be used to translate scientific experiments into a
format that can be interpreted by a computer. The researchers have
published the software code online so that anyone can use and modify it.
Quick comparison

"If lots of scientific papers were written in this way you could very
quickly see whether an experiment has contradictions or agreements with
other work," King told New Scientist. "It would also allow much more
sophisticated search engines to find what you're looking for."

EXPO provides a descriptive framework, or ontology, to represent
different stages of an experiment and the relationships between these
stages. It also includes ways to define the hypothesis tested, the way
results are analysed, and the conclusion drawn.

Researchers already use similar methods to represent specialised
information, such as the genetic properties of different fly species, so
that these can be stored in a computer database and accessed easily.
However, EXPO is the first attempt to apply such an ontological
framework across different sciences.

This means it can be used to compare the experimental methods used in
very different scientific papers. The researchers used two existing
papers to test EXPO – one on particle physics and one on evolutionary
science.
Common ground

Translating the papers using EXPO made it possible for a computer to
pick out similarities that might otherwise go unnoticed. For example, it
revealed that both papers used a method of analysis known as a
"statistical branching model".

"I don't suppose many particle physicists read evolution papers, but I'm
sure both sides have done work that could be useful to the other," King
says. "The maths is very similar – it is just used at very different
time scales."

Ian Horrocks, an information management expert at Manchester University,
UK, who was not involved with the research says it could prove useful in
many practical situations.

"I have been involved with one large pharmaceutical company very
interested in using ontologies to describe the work in its labs," he
says. "They do so much at sites across the world that they often end up
doing the same experiment twice, even in the same lab."

If computers could read experimental write-ups, such duplication would
be avoided, he says: "It could save huge amounts of time, effort and
money. They could avoid repeating what’s been done before and make
better use of the knowledge that already exists."

King admits that for the moment using EXPO is time-consuming because
experimental write-ups must be translated by hand. Software to speed up
this process could be a big boost, he says: "Journals could also insist
that researchers submit papers in EXPO as well as written normally."
Related Articles

 * Can computer models replace animal testing?
 * http://www.newscientisttechnology.com/article/mg19025514.000
 * 16 May 2006
 * Huge protein-interaction database could save lives
 * http://www.newscientisttechnology.com/article/dn8773
 * 24 February 2006
 * Web program simplifies artificial gene design
 * http://www.newscientisttechnology.com/article/dn8737
 * 17 February 2006
 * Computer analysis provides Incan string theory
 * http://www.newscientisttechnology.com/article/dn7835
 * 11 August 2005
 * Particle smasher gets a super-brain
 * http://www.newscientisttechnology.com/article/mg18625005.100
 * 21 May 2005

Weblinks

 * EXPO development site
 * http://sourceforge.net/projects/expo/
 * Computational Biology, University of Aberystwyth
 * http://www.aber.ac.uk/compsci/Department/Research/compBio.php
 * Ian Horrocks research page
 * http://www.cs.man.ac.uk/~horrocks/

http://www.newscientisttech.com/article.ns?id=dn9288&print=true

--
My Original Writing blog - http://itgotworse.blogsource.com


-------------------------------------
You are subscribed as lists-ip () insecure org
To manage your subscription, go to
 http://v2.listbox.com/member/?listname=ip

Archives at: http://www.interesting-people.org/archives/interesting-people/