140 likes | 310 Views
Case-based Reasoning in Design. Mary Lou Maher and Andrés Gómez de silva Garza, University of Sydney IEEE EXPORT,March-April 1997. 指導老師 : 何正信教授 學生:潘立偉 學號: M8702048 日期: 87/10/24. Topic. Learning to design Applications of CBR to design Issue in developing case-based design systems.
E N D
Case-based Reasoning in Design Mary Lou Maher and Andrés Gómez de silva Garza, University of Sydney IEEE EXPORT,March-April 1997 指導老師 : 何正信教授 學生:潘立偉 學號:M8702048 日期:87/10/24
Topic • Learning to design • Applications of CBR to design • Issue in developing case-based design systems
Case-based reasoning is a general paradigm for problem solving based on the recall and reuse of specific experience. Remind designers of previous experience that can help with new situation. What Is Case-based Reasoning?
CBR formalizes a computational model of problem solving based on memory organization and reminding. CBR provides a process model with specific stages. How do Case-Based Reasoning?
Both rely on the explicit symbolic representation of experience-based knowledge to solve a new problem. Expert system use past experience stored in a knowledge base of generalized heuristics to assist in solving a new problem. CBR uses an abstraction of specific experiences to learn to solve a new problem. CBR vs. Expert system
The overall process model of case-based design Design-problem specification Design-case recall Case library Design-case adaptation Design-problem solution
Case recall Design-problem specification Case indexing Design-case recall Case retrieval Case selection To design case adaptation
Case adaptation From design-case adaptation Solution proposal Design-case adaptation Solution evaluation Solution modification Design-problem solution
Complex design cases. Representing a design case as a partonomic hierarchy design cases (Casecad and Cadsyn) partonomic hierarchy of a design case : a decomposition of the design experience into a hierarchy of parts or subcases. Use representation subcases with the Function Behavior Structure (FBS) framework. Function variables Behavior variables Structure variables context variables Applications of CBR to design
representing different view of the same design case (Fable) gestalts (simplified holistic abstraction of a complex geometry) attribute vectors (list of values of predetermined features) images (digital images of the design solution) using multimedia representation for different types of information(Archie-II) photograph, drawing, or stories ...
Generalized design knowledge. Include a representation of general knowledge in the form of causal or heuristic knowledge some methods for representing generalized design knowledge include : causal models, state interactions, heuristic models, heuristic rules , geometric constraints Informal nature of domains.
Representation of design cases and case memory case-memory organization refers to the way cases are organized for access during retrieval. easy : pointer complex : hierarchical indexing tree Issue in developing case-based design system
Classify their variable : structure function behavior Indexing and retrieving design cases. Design often involves solving an ill-defined problem.
Design-case adaptation the human designer a knowledge-intensive computational method a knowledge-lean computational method summary of automated design-case adaptation method