| Article ID: | iaor20061251 |
| Country: | Japan |
| Volume: | 48 |
| Start Page Number: | 66 |
| End Page Number: | 84 |
| Publication Date: | Dec 2005 |
| Journal: | Transactions of the Operations Research Society of Japan |
| Authors: | Mukohara Tsuyoshi, Sekiguchi Yasuki, Bao Jinyuan |
| Keywords: | production, manufacturing industries, practice |
Problems and mathematical models are two essential elements in OR/MS. Particularly, there is some accumulation of research on mathematical model building for decision support problems. By contrast, very few researches have focused on the methodology for structured and operational description of problems. Such description is called problem specification and should be constructed prior to mathematical model building. From this point of view, the authors have proposed GERM as problem specification method. GERM is an extension of the traditional Entity-Relationship Model approach used in designing databases. First, this paper explains the outlines of this extension. In GERM approach, a problem specification is represented as a system type, which consists of a set of object types (entity type and relationship type) and a set of objects (instances of fixed object types and those of open object types). These object types are defined as a set of user defined type attributes. Thus, GERM vocabulary is expandable and flexible. In addition, because type attributes and their values can be some natural terms in a real world of the problem being specified, they are useful for decision makers and model builders to communicate each other. Then, a general form of problem specification by GERM is proposed and it is explained in detail by using the classical F2//Cmax type problem. A description of the problem by the classical entity-relationship approach is compared to a GERM description. Finally, this paper examines and demonstrates the characteristics and the effectiveness of GERM. Because of its features, GERM can deal with not only simple problems like F2//Cmax, but also practical and complex scheduling problems. This is exemplified by describing job shop problems dealt with by a powerful APS. In conclusion, GERM is useful as a constituent for a new architecture of decision support systems which start supporting from problem specification process.