ABSTRACT

In academic research, expert systems have mostly been developed in LISP, although the Japanese and Europeans tend to favor PROLOG. Relatively few programmers outside the research environment are proficient in these languages, however. While LISP and PROLOG offer advantages over procedural languages in their ability to perform symbolic reasoning, it is not necessarily expedient to use these languages in the industrial environment. Neither LISP nor PROLOG are easily interfaced with other languages, so that existing programs, written in FORTRAN or BASIC, would most likely have to be rewritten if they are to be used in conjunction with the expert systems. If expert systems are to be used in the construction or engineering environments, however, it is extremely important to be able to exploit the existence of this base of programs. (This consideration may not be important in other disciplines, where there may not be an existing program base.)

The alternative to using an AI language is to use a commercial expert system shell. The major difference between using such a development tool as opposed to the use of a language is that shells usually contain both an inference mechanism and a knowledge acquisition module, and these do not have to be programmed by the developers. Of course, using a shell limits the structure of the system to its particular inference techniques and representation method, and as such, there is a certain loss of flexibility that is inherent in the bottom-up (language level) approach. In the construction engineering environments, the ability of the expert system to interface with other software is highly significant. In order to provide a useful and complete solution, it may be necessary to use analysis routines, graphics, or data-base management systems. Consideration of such requirements is necessary when selecting appropriate expert system software. Many commercially available shells allow for these interfaces, but some are stand-alone systems, and it is important to recognize the need for an integrated solution before a particular tool is selected.