作者
Warren S. Martin,Warren T. Jones,Evan McWilliams,M. Vernon Nabors
摘要
DEVELOPING ARTIFICIAL INTELLIGENCE APPLICATIONS: A SMALL BUSINESS DEVELOPMENT CENTER CASE STUDY During the last decade, artificial intelligence (AI) programs have emerged as a promising new technology for structuring, guiding, and improving information processing for decision making. AI programs give consultative advice to physicians about infectious diseases; help physicists examine unknown molecules and predict their molecular structures with spectroscopic analysis; assist mathematicians in solving complex problems (Harmon and King 1985); process credit requests for American Express; hunt submarines for the U.S. Navy (Kupfer 1987); help create advertisements for retailers (McCann, Tadlaqui, and Gallagher 1990); and evaluate a client's potential for repaying a loan (Waterman 1985). AI computer software, in some sense, can think. In other words, the programs can problems in a way that would be considered intelligent if done by a (Waterman 1985, 267). AI programs will build human knowledge and processing into an interactive system, draw from that knowledge, and then present selected information that helps solve problems (Nilsson 1980, Van Horn 1986). The subdivision of AI with the most promise for small businesses is systems. Expert systems are programs that contain the information processing ability of experts in a given area. With the proliferation of personal computers and programs for designing AI applications, expert systems are within the reach of small businesses and Small Business Development Centers (SBDCs). While some AI applications are costly and time consuming, potential cost-effective applications for small businesses exist. SBDCs and small businesses need to know the capabilities of expert systems and become familiar with case studies of applications. This article documents an AI application for an information transfer task at an SBDC. THE STRUCTURE OF EXPERT SYSTEMS The major components of expert systems are the knowledge base and the inference engine. The knowledge base is the set of information collected from expert(s) for the structuring, focusing, and processing of information to answer questions. The definition of objects and variables in a chosen area or domain creates the elements of the knowledge base. The relationships between variables and objects are defined and coded as rules. The most common example of a knowledge-base rule is the if-then statement: If a given event occurs, then a given step or question follows. To make the knowledge base compatible with the AI software, each process is reduced to a detailed logical sequence with rules and branching. Since even simple processes have several decision paths, many alternative sets of logic are necessary. Knowledge bases are usually large and complex. The inference engine controls the internal logic relating the facts and rules to the user's information. The direction of analysis can be forward or backward, working from either the beginning logic to a conclusion or from a conclusion to the beginning logic. The inference engine will prompt the computer to ask the user questions; and the answers determine branching to other questions and the outcome. The inference engine controls the AI's speed and style, while the knowledge base controls the AI's content. The result will be expert opinions and information about a given opportunity. DEVELOPMENT OF AN AI APPLICATION Two major elements in the development of AI systems are the selection of the program shell to be used and the process of developing information for the shell. Several companies offer computer programming shells--general frameworks with defined logic paths and rules--for the development of AI applications. The developer uses the logical templates to create the AI application. The shells contain the raw material and a general structure, but the developer needs to supply the specific blueprint, frame the application, and adjust the fit. …