Dear UAI-mailing list members,
For making things clear, let me revise the SCOPE of Chance Discovery I
sent the other day. I appologize for sending seemingly similar mails, but
please note that the content is revised and made clearer.
Thank you
Yukio OHSAWA
Dr. Eng, Associate Professor in University of Tsukuba
3-29-1 Otsuka, Bunkyo-ku, Tokyo 112-0012 Japan
Tel:+81-3-3942-7141, Fax: +81-3-3942-6829
***************************************************************
******** KES 2000, Special Session on Chance Discovery ********
******** 30, 31 August, 1 September 2000, Brighton UK *********
http://luna.bton.ac.uk/~kes2000/
***************************************************************
*** SCOPE ***
- The best way to predict the future is to invent the future (Alan Kay) --
A chance means a new event, which has possibility to become a significant
advantage (damage) in the future life of people, if appropriate efforts
are (are not) made. The difference of discovering chances from predicting
the future is that chances may not really affect the future if people
ignore the chances or avoid the risks. In other words, a chance is an
entrance into the way of inventing or surviving the future, rather than
the future itself - one has to open the door and walk through it.
Two essential points of a chance (or a risk) are:
- A chance is not a sheer repetition of past success, but a trigger
of a new and significant progress. A source of chance is the explosion
of potential motivations, e.g. potential desires of customers in a market.
- New chances are more beneficial than past frequent success-patterns,
because new chances are not known yet by your rivals.
- New risks are more dangerous than past frequent damage-patterns,
because you do not know how to avoid them.
Example 1: When the first rain-umbrella appeared in London, people could
not tell it would become a popular fashion - they rather felt it strange
to walk the street with an umbrella. The difficulty of prediction is
here: there was no time-series of umbrella sales before the first umbrella.
But this was a chance anyway -- thanks to the efforts of advertising
the merit of umbrella "umbrella protects you from rain," it prevailed around
the world. This promotion is a human-information interaction for putting
the chance (umbrella) into a real future fashion. If the umbrella seller
was aware that walking with an umbrella looked strange, he could have found
a chance of selling coats matching with an umbrella.
Here we find it essential to stimulate potential motivations (i.e. the
desire to avoid rain in London, and the sense of fashion of people in London),
for putting a chance into a success. How can we discover the link between
the potential (unknown) motivations and the new product which is unknown,
without any frequent patterns ?
Example 2: When small earthquakes occurred in Kobe and the north of Osaka
(in Japan) from 1980's, people could not tell what they meant. In fact this
was a fatal risk -- due to the stress in the land crust of Kobe, between the
north of Osaka and the trough in the south-east of Japan, 6600 people were
victimized from the big one of M7.2 in Kobe, 1995. If we knew that the past
small earthquakes were causing heavy stress at the focal active fault of
1995, the risk would have been recognized and many people could have been
helped.
Here we find it essential to know the potential causes od risk, for
avoiding a disaster. How can we discover the link between the potential
(unknown) causes and the new disaster which is unknown (an earthquake of
M7.2 is a totally different event from one of M3.0 in the past, even if
both occurred in the same fault which quaked in the past), without any
frequent past patterns ?
For both examples, the key issue is focusing attention to a new event which
is significant for the future, not discarding it as a noise only because it
is unknown. For telling the significance here, human-information interaction
for telling the link between the potential motivations and the new event is
highly contributing to chance discovery.
Many on-going researches including the discovery of
- new products worthy to promote the sales, and
new good customers to send advertising mails, for exploding the sales
- new risks which should be avoided in business and human life
- new keywords in research papers showing pionieering and meaningful
directions of research
- new keywords in WWW which show attracting future trends
and information visualizations, for telling how to put the chance into a real
merit and how to avoid the risk, are very relevant to the session. New
assertions about chance discovery on previous data are welcomed, as well as
new data presenting chances.
*** PAPER SUBMISSIONS ***
The paper dead line is May 15th, by which no more than 4 pages of IEEE
format in
http://luna.bton.ac.uk/~kes2000/#papers
http://luna.bton.ac.uk/~kes2000/guide.htm
is welcomed to be sent to me ***electronically***. The most welcomed style
is a postscript file, gzipped and uuendoded. Please kindly e-mail me
your will to submit and a short paper abstract, before you submit the paper
************************************************************************
** IF YOU HAVE SOME REASON FOR SUBMISSION DELAY, PLEASE INCLUDE IT IN **
** THE MAIL.
************************************************************************
