schultz@grebyn.com (Ronald) (12/14/89)
We are beginning to contemplate developing expert systems for
internal use within our business. Our management has questioned
us on a number of points, and I am respectfully requesting anyone
provide me some assistance based on their own experiences on
answering these questions.
o What are the attributes of a business application
suitable for being incorporated into an expert system ?
o What approach does one take to domain analysis to
identify candidate expert system applications, and how
is this analysis presented to senior manegement to get
buy-in from them for the efforts identified ?
o How does one reduce user expectations associated with all
the hype behind expert systems ?
o How many rules are reasonable to implement into an expert
system running on an IBM PC using software such as M1 or
Level 5 ?
o What attributes does one look for in a potential
knowledge engineer ?
o What success have people had in getting experienced
procedural language programmers with business systems
development expertise in becoming knowledge engineers ?
o What is the application lifecycle for an expert system,
and how long should a successful expert system take to
complete ?
o What techniques can be used to improve the possibility
of success for introducing expert systems into a company
?
Any support, personal experiences, anecdotes, or literature
references to help me address any of the above points would be
sincerely appreciated.
Ron Schultz
Network Solutions, Inc.
Columbus, Ohio
(614) 841-1110
schultz@grebyn.comntm1169@dsacg3.dsac.dla.mil (Mott Given) (12/16/89)
In article <14097@grebyn.com>, schultz@grebyn.com (Ronald) writes: > > We are beginning to contemplate developing expert systems for > internal use within our business. Our management has questioned > us on a number of points, and I am respectfully requesting anyone > provide me some assistance based on their own experiences on > answering these questions. > I would recommend a book by Ed Feigenbaum called something like "The Rise of the Expert Company." It has good descriptions of projects that have worked and ones that have failed. > o How many rules are reasonable to implement into an expert > system running on an IBM PC using software such as M1 or > Level 5 ? You can get around 2000-2500 rules using M1. It depends on the size of the rules. There are ways to get around these limitations by putting some knowledge into an external file, subdividing the knowledge base into separate files, etc. > o What techniques can be used to improve the possibility > of success for introducing expert systems into a company > ? Make sure you get the end users of the expert system application involved from the very beginning! Your system is not going to meet their requirements otherwise, or they will not use it. -- Mott Given @ Defense Logistics Agency ,DSAC-TMP, P.O. Box 1605, Bldg. 27 Section 1, Systems Automation Center, Columbus, OH 43216-5002 INTERNET: mgiven@dsac.dla.mil UUCP: ...{seismo!osu-cis}!dsacg1!mgiven Phone: 614-238-9431 AUTOVON: 850-9431 FAX: 614-238-3214
suji@hpindda.HP.COM (Sujatha Bodapati) (12/20/89)
Another good book to read is 'Crafting knowledge-based systems - Expert systems made realistic' by John R.Walters and Norman R.Nielsen. I was in a similar situation at work and this book helped me come up with a feasability report.