Structured Functional Decomposition Approach to Knowledge-Based Business Process Modeling

This paper presents an approach for structured decomposition of knowledge in certain problem domains, which can be classified as knowledge-based business processes. Knowledge in such problem domains is already partly formalized, and allows natural functional decomposition; however, it is complex enough to justify knowledgebased implementation of process execution. Examples include, in addition to many complex business processes, such areas as law consulting, notary services, etc. The approach is based on structured functional decomposition towards primitive processes that can be described using traditional knowledge representation techniques, such as AND/OR trees or production rules. In this decomposition, both the structure of problem domain concepts and their dynamic properties are identified. A graphical notation is proposed, based on SADT/IDEF-0, extended with knowledge representation notations (graphical as well as textual) to define dynamic properties of lowest-level primitive processes. Process model in this graphical notation can be effectively translated into one-level graphical notation of extended conceptual graphs, as well as into production or production-frame knowledge representation that can be used in creating intelligent automation and/or expert systems. It is demonstrated that forward-chaining logical inference in resulting knowledgebase is modeling business process execution. Some details of implementation are also discussed, and the modeling tool supporting the proposed approach on Microsoft Visio platform is presented. The role of the proposed modeling technique in the overall methodology for knowledge-based systems development is outlined.