Path: utzoo!utgpu!watserv1!ria!csd.uwo.ca!ling From: ling@csd.uwo.ca (Charles X. Ling) Newsgroups: comp.ai Subject: A summary of classification building tools Message-ID: <2364@ria.ccs.uwo.ca> Date: 25 Feb 91 18:22:11 GMT Sender: news@ria.ccs.uwo.ca Organization: University of Western Ontario, London Lines: 152 Originator: ling@csd.uwo.ca Hi, I posted the follwoing request to the comp.ai, and I have received more requests than replies. A summary of what I have got is attached. My original message was: >I am looking for machine learning tools for building up a classification >system. I heard that most successes in inductive construction of >classification systems (and expert systems) are derived from Quinlan's >ID3. If you have used such commercially available packages, >please e-mail me at your earlest convenience: >1. Your experience with the software. The domain you have worked on. > Did the tool help greatly in the system development? >2. Which company sells the package, how much? Their address and/or > phone number. Replies are from: *************************************** From reppenhagen.pad@nixpbe.uucp Thu Feb 21 06:37:19 1991 Received: from unido.informatik.uni-dortmund.de ([129.217.64.60]) by csd.uwo.ca; (id AA09131) Thu, 21 Feb 91 06:36:35 EST From: Date: Thu, 21 Feb 91 11:36:04 GMT Message-Id: <9102211136.AA11331@unido.informatik.uni-dortmund.de> Received: from nixpbe by unido.informatik.uni-dortmund.de with UUCP (5.65+/UNIDO-2.0.3.h) via EUnet for csd.uwo.ca id AA11331; Thu, 21 Feb 91 11:36:04 GMT Subject: commercial knowledge acquisition tool Apparently-To: ling@csd.uwo.ca Status: RO With our experience in real world applications we can recommend NINA-C for automating knowledge acquisition tasks. NINA-C is available for $ 3000 Contact adress: Regina Reppenhagen, Siemens Nixdorf Informationssysteme AG, AI Centre AP333 Pontanusstr. 55, W-4790 Paderborn Email: reppenhagen.pad@sni.de Tel. 49 5251 8-46485, Fax: 49 5251 8-46476 from USA: reppenhagen.pad@sni-usa.com The Induction Tool N I N A - C The C-Induction Tool is a software tool for generating rules from case data efficiently. It is a knowledge acquisition tool, which supports you with the building of expert systems. Overview of Functionality Function ======== Generating rules from examples Features of NINA-C ================== + Extension of the ID3-algorithm + Generating of * Decision trees from classified examples * TWAICE rules (possibly with confidence factors) + Handling of real application data * noisy data (postpruning process) * incomplete data (don't care values) * contradictory data + Data types: real, integer, symbolic, boolean + Selectable split type per attribute (binary tree or value tree) Operators in premises: = <> <= > + Attribute-attribute-comparisons (e.g. in addition to attribute = value, attribute = attribute) + Performance (NINA-C is programmed in C) + Handling large amounts of data + Simple operation * Examples in table form (ASCII file) * Induction trace ******************************* From db@info.ucl.ac.be Thu Feb 21 10:10:03 1991 Received: from mcsun.EU.net ([192.16.202.1]) by csd.uwo.ca; (id AA09862) Thu, 21 Feb 91 10:09:45 EST Received: by mcsun.EU.net via EUnet; id AA14419 (5.65a/CWI-2.74); Thu, 21 Feb 91 16:08:31 +0100 Received: from [130.104.2.23] by ub4b (5.64+/ub4b_01) id AA05569; Thu, 21 Feb 91 15:43:03 +0100 Received: from info-stlz0 Received: by info-stlz0.info.ucl.ac.be (4.1/SMI-4.0) id AA04808; Thu, 21 Feb 91 15:38:06 +0100 Date: Thu, 21 Feb 91 15:38:06 +0100 From: db@info.ucl.ac.be (Daniel Brics) Message-Id: <9102211438.AA04808@info-stlz0.info.ucl.ac.be> To: ling@csd.uwo.ca Subject: classification Status: RO Dear Mr. Ling, In your News mailing, you ask for some information about a good package of classification algorithms. Prof. Quinlan is selling under license a classification package (7000 C lines) for about 80 dollars. Easy to install. Well done. Numerous features. You can contact him by mail at "quinlan@cs.su.oz.au" for more information. Best regards, Daniel Brics. ********************************* From mgv@USCEAST.CS.SCAROLINA.EDU Thu Feb 21 11:02:08 1991 Received: from RELAY.CS.NET ([192.31.103.4]) by csd.uwo.ca; (id AA10148) Thu, 21 Feb 91 11:02:00 EST Received: from oak.cs.scarolina.edu by RELAY.CS.NET id aa08167; 21 Feb 91 9:44 EST Return-Path: Received: by usceast.cs.scarolina.edu (cs.grandpoobah.121990) Date: Thu, 21 Feb 91 09:47:04 -0500 From: Marco Valtorta Message-Id: <9102211447.AA01497@usceast.cs.scarolina.edu> To: ling%csd.uwo.ca@RELAY.CS.NET Subject: Re: Tools for building classification systems Newsgroups: comp.ai References: <2314@ria.ccs.uwo.ca> Status: RO Charles: There are several tools labeled "inductive" in the appendix to Harmon, P. and B. Sawyer. _Creating Expert Systems_, Wiley, 1990. (For example, RuleMaster by Radian Corp.) ...... One tool that is not there is KATE. KATE is (almost--opinions vary) a spinoff of ESPRIT project P1063, INSTIL. It is sold by INTELLISOFT, 28 rue Georges Clemenceau, 91400 ORSAY, FRANCE. Tel: (33)(1) 64-46-28-88 or 45-88-93-02. *************************************** From HOPPET%DB0TUI11.BITNET@tubvm.cs.tu-berlin.de Sat Feb 23 08:44:21 1991 Received: from opal.cs.tu-berlin.de ([130.149.19.1]) by csd.uwo.ca; (id AA24516) Sat, 23 Feb 91 08:43:58 EST Received: from tubvm.cs.tu-berlin.de by opal.cs.tu-berlin.de with SMTP id AA02915 (5.65a/IDA-1.4.2 for ling@csd.uwo.ca); Sat, 23 Feb 91 14:41:59 +0100 Message-Id: <9102231341.AA02915@opal.cs.tu-berlin.de> Received: by DB0TUI11 (Mailer R2.07B) id 8887; Sat, 23 Feb 91 14:42:00 +0200 Date: Sat, 23 Feb 91 14:39:07 +0200 From: Thomas Hoppe Subject: Re: Tools for building classification systems To: ling@csd.uwo.ca Status: RO If your are interested and you would like to experiment, I have an ID3-like implementation in Prolog. Send me a short notice. Best regards, Thomas Hoppe (email: hoppet@db0tui11.bitnet) Brought to you by Super Global Mega Corp .com