Qualitative Business Decision Making

We are developing artiicial intelligence programs that model qualitative business decision making, using both goal-based and case-based reasoning. In this paper, we describe current work in three related areas: normative business goals and beliefs, qualitative business calculus, and belief representation. Overview We are developing artiicial intelligence programs that incorporate goal-based reasoning (GBR) and case-based reasoning (CBR) to simulate decision making in a variety of business domains. In this paper, we provide describe three current projects. We incorporated these features into the VOTE program , which simulated the roll call voting decision making of members of the United States House of Representatives. Given a member of Congress and a speciic bill, VOTE would try to determine how that member would vote and then produce a natural language explanation of the resulting decision in English or French. We have begun to change domains from politics to business. Our programs explicitly represent goals and relationships. This requirement was proposed initially in our description of an advisory system. It is critical that the program examine a problem in the particular context of the present client. The program has to know about the client's goals and needs; what speciic ramiications each alternative may have for the client; how the decision might aaect the client's employees, owners, competitors, customers, and suppliers; and what priorities the client places on these possible eeects. Schank and Slade, 1984, page 251] In extending VOTE to business domains, we have identiied three key research areas. Normative Business Goals and Beliefs. If we want our programs to reason based on goals, we must rst develop a vocabulary or taxonomy of those goals. Business decisions are predicated on a set of implicit goals and beliefs. A computer program must represent this knowledge explicitly. Therefore, we need an inventory of normative business goals and beliefs. We have developed a case-based reasoning tool for cataloging this type of business knowledge. This program may also be used for case analysis and teaching. Qualitative Business Calculus. Though business decisions rely on goals and beliefs, to a great extent such decisions are driven by numbers. Most existing business decision models are quantitative, applying rigorous analytical methods to numeric data. We suggest that qualitative methods can complement the traditional quantitative methods, by providing both an initial justiication for applying a quantitative analysis, and a meaningful interpretation of the quantitative results. We have developed a qualitative business calculus that begins …