Representing modeling knowledge in an intelligent decision support system

摘要

Knowledge representation for data, models, and other decision support system (DSS) elements is a complex and ever-adapting task. The representation scheme for an intelligent DSS will need to provide general problem-solving model management activities as well as a mechanism for refining and testing the applicability of these models for each problem instance it encounters. We present traditional knowledge representation alternatives, and demonstrate why a multi-level scheme is superior for DSS use. We advance a two-level scheme, joining the advantages of connection graphs for the generalized analytical requirements and a frame component for problem-specific query resolution.