| The current theory of inventory management is basically mathematical (quantitative) in nature. However, a large part of inventory management appears to involve judgment and decision making (qualitative), and in the past, little effort has been made to integrate these processes. The objective of this research was to build a prototype expert decision support model (EDSM) using expert systems, natural language, and fuzzy logic technologies to integrate some of the quantitative and qualitative processes of inventory management.;The U.S. Postal Service's material management processes were used to develop the prototype EDSM. Four functions within the inventory management process were identified as those that contained the largest degree of quantitative and qualitative content. These were (1) analysis of historical demand data with large spikes, (2) assessing the reliability of the recorded balance on hand and contract due-in data, (3) determination of purchasing and production lead-times to replenish stock, and (4) anticipating future events that would create unplanned demand.;An expert system shell was used to develop a model reflecting these four functions and their relationships. A set of natural language expressions was developed to permit the inventory specialist personnel to express their degrees of belief relevant to the qualitative decisions, and a fuzzy engine was used to process the natural language expressions.;The prototype EDSM was evaluated using multi-attribute utility technology (MAUT). Three primary dimensions were used in the MAUT evaluation: (1) functionality, (2) usability, and (3) system's fit. The evaluation indicated that the users preferred to use natural language to that of numerical probability statements, had a good degree of confidence in the results, and believed that the system improved their ability to make decisions when dealing with uncertainty in the inventory management process. |
