The ECN No Name Newsletter is no longer being published. This is an archived issue.
[previous article] [next article]The School of Civil Engineering has recently completed negotiations with Software A&E, Inc. resulting in a port of their expert systems shell known as KES to the Gould 9080 under UTX32/2.0. The system has recently been installed on CA and is currently being tested. This package will be used extensively in CE 597d: "Prototyping Expert Systems" Spring semester.
Expert systems are computer-based software systems designed to achieve high levels of performance requiring special algorithms and programming support tools. While conventional software systems operate by applying an algorithm or program to a static data base, expert systems utilize a specialized control structure or "inference engine" that is separated from the "rules set" or "knowledge representation" as stored in a knowledge base. Satisfactory system performance traditionally required the use of programming languages such as LISP, PROLOG or C. Increasingly, however, researchers are making use of knowledge representation languages or "shells" to speed system development and enhance system use.
Modern expert system shells are domain independent systems that allow applications to be designed and developed interactively without the requirement that the designer be proficient in producing computer code. Consequently, designers are able to spend more time and effort on the actual content of the knowledge base than on its structure. In addition, most modern shells provide a wide range of options for knowledge representation and inference engine operation including Bayesian statistics, literal rule-based production systems, and hypothesize-and-test routines. The more complete shells also provide the capability of using external programs and routines such as graphics and data acquisition. The shell may thus be embedded into larger applications.
HTML