--
Mining and Knowledge Discovery in Databases (KDD) community, focusing on
the latest research and applications.
Submissions are most welcome and should be emailed,
with a DESCRIPTIVE subject line (and a URL, when available) to kdd@gte.com
To subscribe, email to kdd-request@gte.com
message with
subscribe kdd-nuggets
in the first line (the rest of the message and subject are ignored).
See http://info.gte.com/~kdd/subscribe.html
for details.
Nuggets frequency is approximately 3-4 times a month.
Back issues of Nuggets, a catalog of S*i*ftware (data mining tools),
and a wealth of other information on Data Mining and Knowledge Discovery
is available at Knowledge Discovery Mine site http://info.gte.com/~kdd
-- Gregory Piatetsky-Shapiro (editor)
********************* Official disclaimer ***********************************
* All opinions expressed herein are those of the writers (or the moderator) *
* and not necessarily of their respective employers (or GTE Laboratories) *
*****************************************************************************
~~~~~~~~~~~~ Quotable Quote ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Arguing with engineers is like mud-wrestling with pigs.
Sooner or later you'll realize that they like it.
Thanks to Tom Lanning Previous1NextTop
>~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
From: 'Edward Colet'(ecolet@watson.ibm.com)
Date: Tue, 11 Feb 1997 16:11:17 -0400
Subject: 'ESPN to regularly show the application of data mining'
On Sunday mornings from 9:00-9:30 (EST), ESPN will regularly broadcast
a show called 'NBA Matchups presented by IBM'. The show will feature
in-depth analysis of player and team match-ups based on trends and patterns
found by Advanced Scout that pertain to the National Basketball Association
(NBA) game of the week. The game of the week is aired later that afternoon
on NBC. Bob Hill (former coach of the San Antonio Spurs), Fred 'Mad Dog'
Carter and Mark Jones (both of ESPN) are the hosts, and an invited guest
will round out the panel (last week's guest was Red Auerbach). As some of you
may know, several NBA coaches have been using IBM's Advanced Scout data
mining application to discover trends and patterns in game data. Advanced
Scout is also the basis for the 'Beyond the Box Score' feature on the
NBA website (www.nba.com. Look under 'News and Features'
if you don't see it off the home page).
Thanks,
Ed Colet.
*********************************************
IBM T.J. Watson Research Center
30 Saw Mill River Road
Hawthorne NY 10532
phone: 914-784-6621; tie-line 863
fax: 914-784-7455
email: ecolet@watson.ibm.com
********************************************* Previous2NextTop
>~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Date: Wed, 5 Feb 1997 10:10:31 -0800
From: datamine@ix.netcom.com
(IDI)
Subject: OLAP & DM Press Release
************************************************************************
Special Release
CONTACT: IDI MARKETING COMMUNICATIONS
(310) 937-3600
Breakthrough Merges
OLAP and DataMining
The Bridge Between OLAP and Data Mining
Impacts all Corporate Decision Support Plans
Los Angeles -- January 27, 1997
The 2nd Annual Data Mining Summit in San Francisco, California on February
19, 1997 is likely to be remembered as the event in which On Line Analytical
Processing (OLAP) and datamining came together for the first time and took
uniform shape.
Up until now, most corporations had considered data mining and OLAP as
individual and disparate components of their decision support system,
because no coherent theory and methodology existed for a relationship. The
1997 Data Mining Summit will bridge this gap and will forever change the way
corporations view and use decision support systems.
At the Keynote Address for the Summit, Dr. Kamran Parsaye, CEO of
Information Discovery, Inc. will introduce a fundamentally new theory and
methodology for connecting OLAP and data mining, showing that they must be
merged in order to avoid incorrect and misleading results during data analysis.
'The bridge between OLAP and data mining is not a luxury but a necessity,'
said Dr. Parsaye. 'OLAP analyses and datamining need to be performed
together if we are to trust the results from either' he added. 'In the early
days of relational databases, before normalization theory was introduced,
people were getting incorrect results. Now, unless OLAP and data mining are
performed together a similar situation can prevail' he said.
The keynote address will show that whenever data analysis takes place, it
happens within some 'dimension', and datamining along a single axis is
merely a rough approximation of multi-dimensional mining. Lack of attention
to dimensionality in data mining can result in unexpected results. And,
decision support errors can take a long time to be uncovered -- if ever. A
companion paper in the February issue of Database Programming and Design
magazine details examples of this phenomena and outlines a uniform approach
for dealing with both OLAP and datamining.
At the keynote, Dr. Parsaye will also describe how OLAP and data mining fit
together in the context of the Four Spaces of Decision Support. This
methodology for applying OLAP data mining has three distinct processes of
episodic, strategic and continuous mining for specific user groups within
corporate environments.
'Integration between OLAP and data mining can not take place at the desktop
level and must be performed on the server' said Dr. Parsaye. 'IS departments
that hand their users OLAP data to be mined on the desktop could be
unknowingly getting their users into serious trouble' he said.
The impact of the new result on corporate planning for decision support and
data warehousing can be significant. Business users and IS departments can
no longer just consider an OLAP product and a separate data mining system
but will need to consider both at once to avoid the pitfalls outlined in the
keynote. This will also accelerate the use of products for both OLAP and
data mining.
For more information on the DataMining Summit please visit http://www.dbsummit.com
on the internet, or call (415) 905 2267. For more
information on Information Discovery, Inc. please visit http://www.datamining.com
on the internet.
[note: any comments from readers on appropriateness of posting
commercial press releases such as above? GPS]
Previous3NextTop
>~~~Publications:~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Date: Mon, 3 Feb 1997 18:56:37 -0500 (EST)
From: Russell Greiner (greiner@scr.siemens.com)
To: ml@ics.uci.edu,
colt@cs.uiuc.edu,
kdd@gte.com,
connectionists@cs.cmu.edu,
ai-stats@watstat.uwaterloo.ca,
dfisher@vuse.vanderbilt.edu,
uai@ghost.CS.ORST.EDU
Subject: CLNL v4 is here!
CC: jose@kreizler.rutgers.edu,
petsche@as1.scr.siemens.com
Content-Length: 363
We are pleased to announce that the book
'Computational Learning Theory and Natural Learning Systems
Volume IV: Making Learning Systems Practical'
(ed. Russell Greiner, Thomas Petsche, and Stephen Jose Hanson)
Previous4NextTop
>~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Date: Tue, 4 Feb 1997 23:23:12 -0800
From: Ronny Kohavi (ronnyk@starry.engr.sgi.com)
Subject: CFP: Special Machine Learning issue on applications of ML
This is a short reminder that the submission deadline for the special
issue of Machine Learning is in a few weeks. For more information, see
Special Issue on
Applications of Machine Learning
and the Knowledge Discovery Process
Guest editors: Ronny Kohavi and Foster Provost
With the explosion in size of business and scientific databases
(VLDBs), the opportunities and pressure to mine the data and make
novel discoveries have increased dramatically. For many problems,
basic statistical summaries are not sufficient and there is a clear
and recognized need for solutions involving a machine learning
component. For example, modern businesses constantly seek to gain
competitive advantage by tailoring actions to different customer
segments and avoiding the trap of targeting the 'average customer.'
This special issue of the journal Machine Learning will be dedicated
to papers describing work in which machine learning technologies have
been applied to solve significant real-world problems. In particular,
it will focus on the application of Machine Learning technology, the
simplifying assumptions that *cannot* be made in a real-world
application, and the processes that are involved in going from the raw
data to the final knowledge that decision makers seek.
_____________________________________________________________________________
Ronny Kohavi and Foster Provost
mljapps@postofc.corp.sgi.com
Previous5NextTop
>~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
From: foster@nynexst.com
Date: Wed, 12 Feb 1997 09:58:29 +0500
Subject: 3 positions at NYNEX S&T
KB system developer positions at NYNEX Science and Technology
The Integrated Network Services Testing & Analysis (INSTA) group at
NYNEX Science & Technology has three openings for knowledge-based (KB)
diagnostic system developers. The group is involved in building
monitoring, testing and diagnostic systems using state of the art AI
technologies for advanced Telecom networks and circuits. The group has
been building systems that support complete testing and diagnosis of
circuits, both from the central office and in the field. Systems
already built and deployed test and diagnose residential telephone
lines and some of the business services. In addition to these, the
group is currently looking at ISDN and broadband services.
The selected candidate would work on one or more of the following
projects:
- Building KB system for assisting field technicians out in the field in
testing and troubleshooting faults in telecomm circuits. The candidate
will also explore complementing this KB with the KB performing
centralized testing from the Central Office.
- Building KB system for automated centralized testing and diagnosis of
Special (Buisness) service circuits.
- Building monitoring, testing, and diagnostic systems for broadband
circuits.
- Building an intelligent interactive assistant to aid testers in
testing and diagnosing circuits.
Suitable candidates must have the following:
===========================================
- Background in Computer Science or Computer Engineering or Electrical
Engineering.
- Experience in all aspects of building knowledge-based systems
including knowledge acquisition, knowledge engineering, domain and
task modeling, testing, validation, and evaluation of the
knowledge-based systems.
- Good understanding of various AI techniques such as model-based
reasoning, case-based reasoning, neural nets, and machine learning.
- Good analytical skills.
- Quick learner - to quickly acquire relevant domain knowledge.
- Good system building experience
Experience with the following would be a plus:
===================================================================
- Knowledge of data-analysis tools (eg: statistical tools)
- Unix, C, C++, LISP, ARTIM, CLIPS...
- Distributed Client server Architectures
- databases, database wharehousing
- Telecomm experience: Operation Support Systems, Residential Lines,
Special services, broadband services, telecomm network and circuit
testing, alarm monitoring etc.
If interested, please mail a hard copy of your resume to:
Yuling Wu
NYNEX Science & Technology
400 Westchester Av.
White Plains, NY 10604
or email the postscript version to:
yuling@nynexst.com
Previous6NextTop
>~~~Positions:~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Date: Fri, 7 Feb 1997 14:21:24 +0200 (EET)
From: Heikki Mannila (mannila@cs.Helsinki.FI)
Subject: Postdoc position in Helsinki: data mining / pattern matching
/ spatial data
Content-Length: 1366
Postdoctoral position in
data mining / pattern matching / spatial data
University of Helsinki
Department of Computer Science
The pattern matching and data mining group in the Department of Computer
Science, University of Helsinki, has an opening for a postdoc researcher
in the areas of data mining, pattern matching, or spatial data.
The research group combines methods from pattern matching, statistics,
and databases to develop methods for the analysis of large data sets.
The group does theoretical and applied research. Currently, special
emphasis is given to work related to bioinformatics and geoinformatics.
The group is one of the leading ones in data mining and string matching.
Applicants should have a recent Ph.D. or equivalent. The appointment
is initially for one year, starting from September 1997.
Applications should contain a curriculum vita, a list of three referees
and a letter addressing the applicant's suitability for the position.
Applications and inquiries may be submitted by email to
Heikki.Mannila@cs.helsinki.fi
or Esko.Ukkonen@cs.helsinki.fi
before February 28, 1997.
Heikki Mannila Esko Ukkonen
Previous7NextTop
>~~~Meetings:~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Date: Wed, 5 Feb 1997 18:21:51 +0000
From: Steve Cartmell (steve@pap.com)
Subject: PA EXPO97 UPDATE
Westminster Central Hall, London, 21-25 April, 1997
The Practical Application EXPO97 brings together four events under one
roof: PAAM97 - The Practical Application of Intelligent Agents and
Multi-Agents; PADD97- The Practical Application of Knowledge Discovery and
Data Mining; PACT97-The Practical Application of Constraint Technology and
PAP97-The Practical Application of Prolog.
PLEASE VISIT OUR RECENTLY UPDATED WEB PAGES FOR FURTHER INFORMATION ON
Tutorials
Invited Talks
Exhibition
Venue
Hotel reservations
Registration
The Practical Application Company
PO Box 137
Blackpool
Lancs FY2 9UN
UK
Tel: +44 (0)1253 358081
Fax: +44 (0)1253 353811
email: info@pap.com
WWW: http://www.demon.co.uk/ar/TPAC/
Previous8NextTop
>~~~Meetings:~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
From: Blaz Zupan (Blaz.Zupan@ijs.si)
Subject: IDAMAP-97: Reminder and brief Second CFP
Date: Wed, 5 Feb 1997 10:35:20 +0100 (MET)
Reminder and brief Second Call for Papers for
IDAMAP-97
INTELLIGENT DATA ANALYSIS IN MEDICINE AND PHARMACOLOGY
Saturday, August 23, 1997
Workshop W15 at IJCAI-97
August 23-29, 1997, Nagoya, Japan
Paper submission deadline is March 3, 1997. Submit 8-12 page papers by
e-mail (postscript) and 3 hard-copies by surface mail to:
Nada Lavrac, Blaz Zupan
J. Stefan Institute, Jamova 39, SI-1000 Ljubljana, Slovenia
email: idamap97@ijs.si
For up-to-date workshop information please check: http://www-ai.ijs.si/ailab/activities/idamap97.html Previous9NextTop
>~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Date: Mon, 10 Feb 1997 18:00:38 +0100 (MET)
From: Gerhard Widmer (gerhard@ai.univie.ac.at)
Subject: ECML'97 - Papers & Registration Info
-------------------------------------------------------------------------
NINTH EUROPEAN CONFERENCE ON MACHINE LEARNING (ECML-97)
Prague, Czech Republic, April 23-26 1997
******************************************************
ECML'97: LIST OF ACCEPTED PAPERS and REGISTRATION INFO
******************************************************
-------------------------------------------------------------------------
The list of accepted papers, INCLUDING ALL ABSTRACTS, is now
available from the ECML-97 WWW home page:
This page also gives access to
- the 4 post-conference ECML/MLNet WORKSHOPS and
- ECML-97 REGISTRATION INFORMATION and the ECML REGISTRATION FORM.
- A preliminary version of the CONFERENCE PROGRAMME will be available soon.
For further questions about the program, contact Gerhard Widmer
at gerhard@ai.univie.ac.at,
for questions regarding registration,
contact the local organizers at actionm@cuni.cz.
For those without access to the WWW, please find below
- titles and contact addresses for the 4 MLNet workshops,
- the list of papers (w/o abstracts),
- an ascii version of the registration form.
WS 1: Data-Driven Learning of Natural Language Processing Tasks
Contact: Walter Daelemans,
P.O. BOX 90153, NL-5000 LE Tilburg, The Netherlands.
Tel: +31 13 4663070, Fax: +31 13 4663110,
E-mail: walter.daelemans@kub.nl
WS1 WWW Page: http://www.cs.unimaas.nl/ecml97/
WS 2: Case-Based Learning: Beyond Classification of Feature Vectors
Contact: Dietrich Wettschereck,
GMD, FIT.KI, Schloss Birlinghoven,
53754 Sankt Augustin, Germany
Tel: +49-2241-14-2097, Fax: +49-2241-14-2072,
E-mail: dietrich.wettschereck@gmd.de
WS2 WWW Page: http://nathan.gmd.de/persons/dietrich.wettschereck/ecmlws.html
WS 3: Learning in Dynamically Changing Domains:
Theory Revision and Context Dependence Issues
Contact: Gholamreza Nakhaeizadeh,
Research Center of Damiler-Benz AG, Ulm, Germany
E-mail: nakhaeizadeh@dbag.ulm.DaimlerBenz.COM
WS3 WWW Page: http://www.amsta.leeds.ac.uk/statistics/ecml97/dyn.htm
WS 4: Machine Learning and Human-Agent Interaction
Contact: Michael Kaiser,
Institute for Real-Time Computer Systems & Robotics
University of Karlsruhe, Kaiserstrasse 12,
D-76128 Karlsruhe, Germany
E-Mail: kaiser@ira.uka.de
WS4 WWW Page: http://wwwipr.ira.uka.de/events/hai97/
Common dates for all workshops:
Deadline for submissions: February 15
Notification of acceptance: March 8
Camera-ready copy due: April 1
Learning Complex Probabilistic Models (tentative title)
Stuart J. Russell, University of California, Berkeley, USA
Constructing and Sharing Perceptual Distinctions
Luc Steels, Free University of Brussels (VUB) and
Sony Computer Science Laboratory, Paris
On Prediction by Data Compression
Paul Vitanyi, CWI, Amsterdam
Ming Li, City University of Hong Kong
LONG TALKS/PAPERS:
Induction of Feature Terms with INDIE
Eva Armengol & Enric Plaza, IIIA, Barcelona, Spain
Integrated Learning and Planning Based on Truncating Temporal Differences
Pawel Cichosz, Warsaw University of Technology, Warsaw, Poland
Theta-subsumption for Structural Matching
Luc De Raedt, Katholieke Universiteit Leuven, Belgium
Peter Idestam-Almquist, Stockholm University, Sweden
Gunther Sablon, Katholieke Universiteit Leuven, Belgium
Constructing Intermediate Concepts by Decomposition of Real Functions
Janez Demsar, Blaz Zupan, Marko Bohanec, Ivan Bratko
University of Ljubljana and Jozef Stefan Institute, Ljubljana, Slovenia
Conditions for Occam's Razor Applicability and Noise Elimination
Dragan Gamberger, Rudjer Boskovic Institute, Zagreb, Croatia
Nada Lavrac, Jozef Stefan Institute, Ljubljana, Slovenia
Learning Different Types of New Attributes by Combining
the Neural Network and Iterative Attribute Construction
Yuh-Jyh Hu, University of California, Irvine, USA
Finite-Element Methods with Local Triangulation Refinement
for Continuous Reinforcement Learning Problems
Remi Munos, CEMAGREF, Antony, France
Compression-based Pruning of Decision Lists
Bernhard Pfahringer, University of Waikato, New Zealand
NeuroLinear: A System for Extracting Oblique Decision Rules
from Neural Networks
Rudy Setiono & Huan Liu, National University of Singapore
Model Combination in the Multiple-data-batches Scenario
Kai Ming Ting, University of Waikato, New Zealand
Boon Toh Low, Chinese University of Hong Kong
Natural Ideal Operators in Inductive Logic Programming
Fabien Torre & Celine Rouveirol, LRI, Paris, France
Ibots Learn Genuine Team Solutions
Cristina Versino & Luca Maria Gambardella, IDSIA, Switzerland
Global Data Analysis and the Fragmentation Problem in Decision Tree Induction
Ricardo Vilalta, Gunnar Blix, Larry Rendell,
University of Illinois at Urbana-Champaign, USA
SHORT TALK/PAPERS:
Exploiting Qualitative Knowledge to Enhance Skill Acquisition
Cristina Baroglio, Universita di Torino, Italy
Classification by Voting Feature Intervals
G'ulsen Demir'oz & H. Altay G'uvenir,
Bilkent University, Ankara, Turkey
Metrics on Terms and Clauses
Alan Hutchinson, King's College, London, UK
Learning When Negative Examples Abound
Miroslav Kubat, Robert Holte, Stan Matwin, University of Ottawa, Canada
A Model for Generalization based on Confirmatory Induction
Nicolas Lachiche, INRIA Looraine, France
Pierre Marquis, Universite d'Artois, France
Learning Linear Constraints in Inductive Logic Programming
Lionel Martin & Christel Vrain, Universite d'Orleans, France
Inductive Genetic Programming with Decision Trees
Nikolay Nikolaev, American University in Bulgaria
Vanyo Slavov, New Bulgarian University, Sofia, Bulgaria
Parallel and Distributed Search for Structure in Multivariate Time Series
Tim Oates, Matthew Schmill, Paul Cohen
University of Massachusetts, Amherst, USA
Probabilistic Incremental Program Evolution: Stochastic Search
Through Program Space
Rafal Salustowicz & J'urgen Schmidhuber, IDSIA, Switzerland
The GRG Knowledge Discovery System:
Design Principles and Architectural Overview
Ning Shan, Macro International Inc., Calverton, MD, USA
Howard Hamilton & Nick Cercone, University of Regina, Canada
Learning and Exploitation do not Conflict under Minimax Optimality
Csaba Szepesvari, University of Szeged, Hungary
Search-based Class Discretization
Luis Torgo & Joao Gama, University of Porto, Portugal
A Case Study in Loyalty and Satisfaction Research
Koen Vanhoof, Josee Bloemer, K. Pauwels
Limburgs Universitair Centrum, Belgium
TO BE FAXED (42-2) 6731 0503 OR MAILED Action M Agency,
note, please Vrsovicka 68
that after March 1, 1997, 101 00 - Praha 10
the country number (42) Czech Republic
will be changed to (420)
FILL IN CAPITAL LETTERS, PLEASE
last name: first name:
Prof./Dr./Mr./Ms. affilliation:
university/dept.:
street: town:
Code: country:
phone: fax:
e-mail:
name of accompanying person(s):
date (time) of arrival: date of departure:
number of nights:
I will attend workshop: 1. 2. 3. 4. (tick, please)
ACCOMMODATION: krystal hotel (Conference site)
an individual choice up to price per night:
Room: single double NAME OF PERSON SHARING THE ROOM:
special needs (vegetarian, disabled, etc.):
CONFERENCE FEES: BEFORE / AFTER FEBRUARY 20, 1997
CONFERENCE FEE (APRIL 23-25) DM 270.00 / 320.00
MLNet WORKSHOP FEE (APRIL 26) DM 35.00 / 35.00
ACCOMPANYING PERSON FEE DM 80.00 / 100.00
ACCOMMODATION DEPOSIT: DM 150. 00
ACCOMMODATION BALANCE:
(NUMBER OF NIGHTS MINUS THE DEPOSIT )
SOCIAL PROGRAM:
SIGHTSEEING TOUR OF PRAGUE DM 25. 00
TA FANTASTIKA THEATRE DM 27. 00
TRIP & FAREWELL PARTY DM 65. 00
TOTAL AMOUNT:
PAYMENT BY CREDIT CARD:
AMEX VISA Master Card / Eurocard JCB Diners
club
Number:
Knowledge Discovery Nuggets 97:06