Pulter, Daniel S. (author), Zilberman, David (author), and Pulter: Agricultural Economist, Commodity Economics Division, Economic Research Service, U.S. Department of Agriculture; Zilberman: Professor, Department of Agricultural and Resource Economics, University of California, Berkeley
Format:
Journal article
Publication Date:
1988
Published:
USA
Location:
Agricultural Communications Documentation Center, Funk Library, University of Illinois Box: 72 Document Number: C03369
AGRICOLA IND 90036799, Agricultural Expert Systems (AES) are gaining recognition as technology transfer devices. It is increasingly clear that lack of uniformity in their presentation to the user creates unnecessary obstacles to their acceptance. Some agronomic expert system applications require performance and economy not currently available with commercial expert system development shells. A group of expert system developers resolved to establish uniform standards for the part of an AES a user must interact with. A committee was appointed to draft the standards from their collective experience. Software was developed to implement these standards in expert systems, without sacrificing performance and economy. Two lists of standards were developed: mandatory and desirable. These are minimal features, and extended or enhanced features, respectively. The lists were compiled and refined through study of existing programs, current definitions, developers' experience, and user input. Software to meet the standards was developed in C language, in conjunction with an expert system. This permitted interactive refinement of both the standards and the software. The standards for AES features were endorsed by two groups of cooperating AES developers. Public domain source code is available to implement the features mentioned above, in the C programming language. Concurrence upon the two lists of standards proved to be possible. Adherence to such standards is desirable, since it reduces user confusion. Including these features in programs not usually thought of as expert systems, is easy, inexpensive, and greatly increases their acceptability to new users. (original)