News

Software Program Provides Early Warning of Potential Auto Safety Issues By
Automatically Finding Hidden  Patterns In Federal Data of Fatal Crashes.

Findings for Ford Explorer Made Available

Hartsdale, NY August 23, 2001 /PRNewswire/ - Virtual Gold, Inc., a leading
next-generation data mining software company announced today the
availability of a computer program (CarMiner) that will enable users to test
their theories and evaluate information about vehicles involved in fatal
crashes. If a theory presented to CarMiner is incomplete or inaccurate, the
program automatically highlights statistics in the FARS (Fatality Analysis
Research System) data that give the user a more accurate picture.

CarMiner is based on a new approach to data mining called causative
analysis, the latest brainchild of Dr. Inderpal Bhandari, Founder & CEO of
Virtual Gold. The approach involves the use of a computer model of what
statistical patterns make sense from the standpoint of physical causality
and real-world impact.

Dr. Bhandari explained the benefits. "It turns out that using a computer
model of causality makes it easier to assess the impact of the trillions of
partial and conditional relationships that exist among variables in
high-dimensional data. That, in turn, makes it possible to automatically
evaluate very large numbers of patterns efficiently, since most can be
eliminated rapidly from consideration. Finally, it makes it easy for
non-technical decision makers to use powerful analysis methods since
everyone has a commonsense idea of what causality and real-world impact are.
I expect that such models will be used routinely by the next generation of
data mining software". Causative analysis is now an integral part of Virtual
Gold's VirtualMiner technology platform that was used to develop CarMiner.

Virtual Gold also announced the availability of a report on the experience
of using CarMiner to study the Ford Explorer. The report addresses questions
of current interest pertaining to tire-related, rollover-related,
driver-maneuver-related, and in-car-death- related fatal crashes involving
Ford Explorer and other comparable sport utility vehicle (SUV) during the
period 1995-1999. By examining the relationships that exist among 120
vehicle-related and driver-related attributes in the FARS data of over
10,000 fatal crashes, the FARS statistics highlighted automatically by
CarMiner provide insight over and beyond what appears to be presently known
about these questions. They can serve to provide early warnings of potential
safety issues that should be addressed.

Virtual Gold estimated that there could be as many as 344 trillion partial
relationships among the 120 attributes in the FARS data on SUV's, suggesting
why it was all but inevitable that CarMiner would discover knowledge to shed
light on these questions about Ford Explorer. CarMiner processes this data
efficiently on a standard personal computer, which is suggestive of the
power of causative analysis. The report introduces the concepts that
underlie causative analysis and is written for a wide, non-technical
audience. The company also plans to conduct analyses on other models of
SUVs, cars and trucks in the near future.

Dr. Bhandari, Virtual Gold's Founder and CEO, was prompted to apply the
company's award winning data mining technology to the FARS database after
reading an article in the New York Times. The article noted that "examining
the (FARS) data without a clear idea of what it contains is much harder than
finding clues once a problem is known", explaining the challenge faced by
Ford, Firestone and federal safety regulators in identifying such problems
before they are known.

CarMiner is intended for automobile makers, insurance companies, banks, U.S.
and non- U.S. government agencies, not-for-profit organizations, automobile
consumers, and web sites for automobile consumers. Customers can use
CarMiner as an Internet-based service to test their own theories and
evaluate information about vehicles in the FARS or other databases. CarMiner
is also available as a software platform. Details may be found at
www.virtualgold.com/CarMiner/. Parties interested in business development
opportunities should contact Thomas J. Martin, thomas.martin@virtualgold.com
Executive Vice President, who is responsible for taking Virtual Gold's next
generation data mining software to market.

About Virtual Gold, Inc., www.virtualgold.com

Virtual Gold, Inc. is a leading next-generation business intelligence and
data mining software company. Former IBM scientist Dr. Inderpal Bhandari,
one of the world's leading authorities on data mining and its timely
application to real world problems, founded it in 1997. Virtual Gold
provides an integrated family of web-based data mining software and
solutions. The company is headquartered in Hartsdale, New York and has a
software development group located in Kolkata, India.

Source: Virtual Gold, Inc.
Website: www.virtualgold.com