Case‐Based Reasoning Systems in Purchasing: Applications and Development

In today's intensely competitive and turbulent global market, improving the quality of critical purchasing management decisions will result in significant increases in corporate profitability.[1] However, improving decision quality in this dynamic, time-dependent purchasing environment is a challenge. Many purchasing management decisions fall into one or both of the following categories: * Complex - Decisions involve a myriad of alternatives or variables to consider (e.g., purchase of expensive, highly technical products or supplier selection). * Unstructured - Decisions involve variables that cannot be quantified and require management judgment for their solution (e.g., price negotiations or forecasting future material prices). One methodology for improving complex or unstructured purchasing management decision making involves the use of artificial intelligence (AI) computer decision support software. For example, expert systems, which solve problems by emulating the problem-solving behavior of a human expert, have been applied successfully to some areas of purchasing decision making.[2] Expert systems offer significant potential for further application to purchasing decision making but are not applicable in decision situations where no expert exists or where the expertise (knowledge) is not well developed (e.g., knowledge is incomplete, uncertain or inconsistent).[3] Examples of such decision situations include: * planning initial purchase of new expensive technical assets * planning initial price negotiations with a potential major supplier * developing an initial forecast of material prices for a new material * developing an initial sourcing strategy in a foreign country However, a new AI computer software technology, case-based reasoning (CBR) systems, shows significant promise for improving the effectiveness and efficiency of purchasing management decision making in these situations.[4] This article describes CBR systems technology, reports the current status of purchasing CBR system applications, presents several purchasing CBR system applications that offer significant potential for future development, and provides guidelines for CBR system development. CBR SYSTEMS TECHNOLOGY Case-based reasoning systems are computer programs that solve problems by using the problem-solving experiences of humans.[5] These systems are developed by knowledge engineers who interview one or more managers to catalog their experiences (i.e., problems faced and corresponding solutions) regarding a specific problem domain and then construct a computer program using programming languages or CBR system software development tools. CBR systems consist of a knowledge base and an inference engine [ILLUSTRATION FOR FIGURE 1 OMITTED]. The knowledge base contains a library of classified (indexed) problem-solving experiences called "cases." Each case contains information regarding a specific problem situation and its solution. The inference engineer mirrors the problem-solving approach taken by a manager who solves current problems using past experiences. CBR systems provide decision support to managers through an interactive question and answer session. Managers provide input in the form of a description of a current problem and situation. Then, the CBR system classifies the current case and retrieves a case or cases from the knowledge base that are identical or similar to the current problem. If an identical match is found, the matching case and solution are provided. If a similar case if found, the CBR system modifies the case solution to fit the problem and situation; evaluates the proposed solution and, if necessary, revises the proposed solution using management input; indexes this new case (current problem and its proposed solution), stores it in the knowledge base, and updates the index. Then, the CBR system provides the manager with a solution to the problem and the cases that are similar to the situation. …