jej@chinet.chi.il.us (joe jesson) (07/02/90)
I work for a large company and recently been reviewing KA packages for commercial use - what a joke. The maturity of such tools leave a great deal to be desired and have been an academic exercise. The Knowledge-based tools include Acquinas, Salt (if you want to hire Boeing Consultants, fine, if not forget it), etc. Most of the tools discussed by Gaines, Boose, et. al. have never stood up to commercial use and is generally useless as a development tool. For cetain classes of problems (diagnostic), decision trees are of practical use and induction tools represent a _starting_ point. ..
sticklen@cps.msu.edu (Jon Sticklen) (07/02/90)
In <5993@uklirb.informatik.uni-kl.de> jej@chinet.chi.il.us (joe jesson) writes: > > I work for a large company and recently been reviewing KA packages for > commercial use - what a joke. The maturity of such tools leave a great deal > to be desired and have been an academic exercise. The Knowledge-based tools > include Acquinas, Salt (if you want to hire Boeing Consultants, fine, if not > forget it), etc. > > Most of the tools discussed by Gaines, Boose, et. al. have never stood up to > commercial use and is generally useless as a development tool. For cetain > classes of problems (diagnostic), decision trees are of practical use and > induction tools represent a _starting_ point. .. it would be really useful to the community if the shortcomings could be enumerated. ie, what things did you need that current tools did not give you? ---jon---
ahlenius@relay.eu.net (Mark Ahlenius) (07/03/90)
jej@chinet.chi.il.us (joe jesson) writes: > I work for a large company and recently been reviewing KA packages for >commercial use - what a joke. The maturity of such tools leave a great deal >to be desired and have been an academic exercise. The Knowledge-based tools >include Acquinas, Salt (if you want to hire Boeing Consultants, fine, if not >forget it), etc. >Most of the tools discussed by Gaines, Boose, et. al. have never stood up to >commercial use and is generally useless as a development tool. For cetain >classes of problems (diagnostic), decision trees are of practical use and >induction tools represent a _starting_ point. .. Out of curiousity, have you looked at (or used) Nexpert, by Neuron Data (actually by Boose or Gaines, I cant remember which one is at Calgary). Like to know your impressions. I have played with it, and find it interested - but I didn't come away being awe-struck or anything. I have not seen (nor have I looked) for other tools like it. It does help analyze information more easily. I would think its strong points are with classification type problems - do you have any opinions on this area? thanks 'mark -- =============== regards 'mark ============================================= Mark Ahlenius voice:(708)-632-5346 email: uunet!motcid!ahleniusm Motorola Inc. fax: (708)-632-2413 Arlington, Hts. IL, USA 60004
jstout@atc.boeing.com (Jeffrey Stout) (07/04/90)
In article <5993@uklirb.informatik.uni-kl.de> jej@chinet.chi.il.us (joe jesson) writes: > > I work for a large company and recently been reviewing KA packages for >commercial use - what a joke. The maturity of such tools leave a great deal >to be desired and have been an academic exercise. The Knowledge-based tools >include Acquinas, Salt (if you want to hire Boeing Consultants, fine, if not >forget it), etc. > >Most of the tools discussed by Gaines, Boose, et. al. have never stood up to >commercial use and is generally useless as a development tool. For cetain >classes of problems (diagnostic), decision trees are of practical use and >induction tools represent a _starting_ point. .. As one of the people who worked on the development of SALT at CMU (and a Boeing employee as well), I feel an obligation to point out that SALT was in fact used in a commercial application (at CMU). However, you _do_ have a point. This technology is _not_ very mature, especially in the area of constructive systems (such as SALT). My experience has been that the only people who claim the technology is mature are the people who are selling such tools. My work is in the area of knowledge acquisition for scheduling domains, and more than once I've run across people who believe it is a solved problem because they've been talking to a commercial vendor. Commercial KA tools may be "jokes", but it is because they give up a great deal of power by being so general (in order to be easier to market). I don't think anyone would argue that KA tools with a more limited scope tend to be better for the things they do then the general "products" out there. The problem is that these systems do not make much of a dent in "the set of all problems in the world". But we're working on it. Disclaimers: SALT, Silica, and Aquinas are _NOT_ Boeing products. Any opinions expressed here are my own, not Boeing's. "We shall reach greater and greater platitudes of achievement." --Chicago mayor Richard J. Daley =-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-= Jeff Electrons: jstout@atc.boeing.com Personifiers unite! Stout uw-beaver!bcsaic!jstout You have nothing to Molecules: Advanced Technology Center lose but Mr. Dignity! Boeing Computer Services, MS 7L-64 P.O.Box 24346, Seattle WA 98124-0346 "This quote no verb."