KDD Nugget 94:16, e-mailed 94-08-29 Contents: * Tim Hume, UCI ML info available via WWW * Andy Pryke, article on XpertRule Analyzer for credit analysis * Ibrahim Imam, CFP: Intelligent Adaptive Systems (IAS-95) * John Major, Free (almost) time series data available * R. Zicari, General OODB for Software Engineering Processes * Jack Park, Query: multidimensional databases ? The KDD Nuggets is a moderated list for the exchange of information relevant to Knowledge Discovery in Databases (KDD, also known as Data Mining), e.g. application descriptions, conference announcements, tool reviews, information requests, interesting ideas, clever opinions, etc. It has been coming out about every two-three weeks, depending on the quantity and urgency of submissions.. Back issues of nuggets, a catalog of data mining tools, useful references, FAQ, and other KDD-related information are now available at Knowledge Discovery Mine, URL http://info.gte.com/~kdd/ or by anonymous ftp to ftp.gte.com, cd /pub/kdd, get README E-mail contributions to kdd@gte.com Add/delete requests to kdd-request@gte.com -- Gregory Piatetsky-Shapiro (moderator) ********************* 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 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ Why do romances between agents never work out? Because they treat each other like objects. -- AAAI attendee. [David Joslin (joslin@cs.pitt.edu), 8/8/94.] ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ----------------------------- Subject: UCI - ML - WWW Date: Wed, 27 Jul 1994 16:04:19 -0700 From: Tim Hume ****************************************** ** University of California, Irvine ** ** ** ** Machine Learning ** ** ** ** now available through WWW ** ****************************************** The Machine Learning group of the Department of Information and Computer Science at the University of California - Irvine now has information available via WWW. The URL is http://www.ics.uci.edu/AI/ML/Machine-Learning.html From here you can access UCI's repository of databases for machine learning research, digests of the Machine Learning List, programs (FOCL, Occam, and HYDRA) developed at UCI, and papers by authors from UCI. The Machine Learning List digests are searchable, and we plan to have the repository searchable later this year. Comments and suggestions for improvements are welcome. Assistance in making the suggested improvements is also welcome. Please send correspondence to Tim Hume hume@ics.uci.edu UC, Irvine ----------------------------- Date: Thu, 11 Aug 94 15:25:50 BST From: A.N.Pryke@computer-science.birmingham.ac.uk Subject: Summary of Article on an application of KDD Hi, I thought you might be interested in the following summary of an article (perhaps for KDD-Nugget). The package refered to is "XpertRule Analyser" which I assume is a different package to the "XPERTRule" in your Siftware list. Unfortunately I have no further information on the software or the company which supply it. Andy --- Andy Pryke, Research Student, Computer Science, Birmingham University Data Mining Information Summary of @Article{ expert:getting, key = {Expert Systems}, title = {Getting to grips with arrears: `data mining' systems at the {L}eeds}, journal = {Expert Systems}, year = {1994}, volume = {11}, number = {2}, pages = {122--124}, month = may, keywords = {Applications, Data mining, KDD, Attar Software, Xpert Rule Analyser}, } The Leeds Permanent is Britain's 5th biggest building society, with 500,000 mortgage accounts, and has had to make dramatically increasing provision for bad debt over the past 5 years. Credit scorecard techniques were introduced in the early 1990s but could not take account of changes in the circumstances of mortgage holders. More recently, the firm has started using XpertRule Analyser for Attar Software. Attar software and Price Waterhouse work together on building society projects. The software produces regression and decision trees. It handles noise by pre-processing the data, and pruning the derived trees. The system uses genetic algorithm based rule induction to generate multiple decision trees. The article contrasts this (symbolic) approach with the black box approach of neural networks. The software is used to generate sets of rules which profile account holders. Understanding the patterns of risk is essential for generating a risk management strategy. The software has (re)discovered classical risk factors, but also identified new and surprising ones. For example, high-income earners and businesses which had been introduced by a third party. The article ends by predicting "a healthy future for customer profiling technologies such as those offered by Attar Software" No(0) references. The address of Attar Software is not given. ----------------------------- Date: Mon, 15 Aug 94 13:34:36 EDT From: iimam@aic.gmu.edu (Ibrahim Fahmi Imam) Subject: CFP: Intelligent Adaptive Systems (IAS-95) CALL FOR PAPERS FLAIRS-95 International Workshop on Intelligent Adaptive Systems (IAS-95) April 26, 1995, Melbourne Beach, Florida Recently, there has been significant interest in developing intelligent systems or agents capable of dynamically building and adapting their knowledge-base with experience to accomplish a given task. Building such agents is of great practical interest in view of the increased demand for the development of knowledge-based systems, (e.g. personal assistants, intelligent agents, information filters). However, very often the forms in which it is easy to represent the obtained knowledge-base differ from those which are dynamically adjustable for performing different tasks. Also, efficient methodologies for acquiring knowledge-bases must be flexible enough for dynamic adaptation. The goal of this workshop is to bring together researchers working on the development of fundamental strategies and methodologies needed for the development of intelligent adaptive systems. The workshop seeks high quality submissions in which there are links between the presented work, the workshop topic, and potential applications. Papers with real-world applications are very welcome. The areas of interest of the workshop include, but are not limited to: - Multistrategy Learning for Intelligent Agents - Knowledge Discovery Systems - Constructive Induction and Representation Space Improvement - Adaptive Control Systems - Implementations of Intelligent Agents - Knowledge Acquisition via Inductive Logic Programming Program Committee Chairs Imam, I.F. George Mason University, USA Wnek, J. George Mason University, USA Program Committee Members Gordon, D. Naval Research Laboratory, USA Hamilton, H. University of Regina, Canada Michalski, R.S. George Mason University, USA Morik, K. University of Dortmund, Germany Pazzani, M. University of California Irvine, USA Rafea, A. Cairo University, Egypt Ram, A. Georgia Institute of Technology, USA Saitta, L. University of Trento, Italy Schum, D. George Mason University, USA Segen, J. AT&T Bell Laboratories, USA Whitehall, B.L. United Technologies Research Center, USA Ziarko, W. University of Regina, Canada SUBMISSIONS Paper submissions should not exceed ten single-spaced pages, with 1 inch margins, 12pt font. The cover page must show author(s) full address, email and include abstract (200 words maximum) and keywords (5 maximum). Three copies of each paper should be sent to the contact address below. Alternatively, one copy of a postscript file may be sent via email. The papers will comprise a set of working notes (proceedings), copies of which will be available at the workshop. SCHEDULE Paper submissions due December 1, 1994 Acceptance notice January 31, 1995 Camera ready version March 1, 1995 Workshop April 26, 1995 CONTACT ADDRESS Ibrahim F. Imam Machine Learning and Inference Center George Mason University 4400 University Dr. Fairfax, VA 22030 e-mail: iimam@aic.gmu.edu Tel: (703) 993-1716 Fax: (703) 993-3729 ----------------------------- From: jmajor@nyx.cs.du.edu (john major) Subject: Scanner Data Date: Mon, 15 Aug 1994 06:24:09 -0600 (MDT) The following posting appeared as misc.invest.technical #5497. It may be of interest to some KDD'ers. ------------------------------------------------------------------------ From: leobueno@gate.net (Leo Bueno) [1] Free (almost) time series data available Date: Sat Aug 13 18:14:48 MDT 1994 Lines: 16 X-Newsreader: TIN [version 1.2 PL2] From MJWOLFE@delphi.com Sun Aug 7 15:41:24 1994 My company, Kraft General Foods can make a large "UPC SCanner" data set available to academics and researchers intrerested in non- commercial applications of this data...there will be anominal cost to replicate the tape and mailing, but that is considerably cheaper than the commercial rate on tihis tyhpe of data..Mike Wolfe My number (fax) is (708) 646-2454 if interested...mail a requiest.. -- Leo Bueno ====> leobueno@gate.net ------------------------------------------------------------------------ John A. Major, ASA 517 Church Street Research Consulting Newington, CT USA 06110 jmajor@nyx.cs.du.edu (203) 666-7023 ------------------------------------------------------------------------ ----------------------------- From: zicari@informatik.uni-frankfurt.de Subject: Goodstep Date: Fri, 19 Aug 1994 23:48:49 +0200 (MESZ) General Object Oriented Database for Software Engineering Processes (GOODSTEP) ESPRIT-III Project. The objective of GOODSTEP is to enhance and improve the functionality of the O2 object-oriented database system to yield a platform suited to applications such as software engineering environments (SEEs). A number of advanced software engineeiring tools are being built on top of the enchanced O2 database. Goodstep started September 1992 is a three years-project, with a budget of 5,8 MECU and a total of 44 man/year. Goodstep reports are periodically stored at INRIA ftp server. You can obtain GoodStep reports via anonymous ftp from ftp.inria.fr. Login with user name anonymous and cd to INRIA/Projects/Verso/GoodStep-Library. There you can find all compressed postscript files of the reports and an INDEX listing all reports in that directory, with the title, author(s) and additional information, if any. Roberto Zicari Goodstep Technical Director ----------------------------- Date: Sat, 27 Aug 1994 15:37:49 -0700 From: jackpark@netcom.com (Jack Park) Subject: A Question Jeffrey P. Stamen has written a paper "Structuring databases for analysis" in IEEE Spectrum October, 1993 (pp 55-58) in which he compares the multidimensional the relational model in light of analysis. Is any reader on this usenet thinking about or using multidimensional databases? Jack Park