An international conference connecting people
in CAD research, education and business
Bookmark and Share
Copyright (C) CAD Solutions, LLC. All rights reserved.
Proceedings of CAD'16, 2016, 157-161
Reconstructing Design Processes by Machine Learning of Graph-Rewriting Production Rules

Julian R. Eichhoff, Felix, Baumann, Dieter Roller, University of Stuttgart

Abstract. Graph-based models play an important role in product design. Particularly in conceptual design, graphs are used for abstract representation of functionality, topology and physical relations among product components. Graph-rewriting is an expressive computation model operating on graphs, and thus becomes a natural choice for implementing computed-aided conceptual design. Graph-rewriting systems rely on a set of so-called production rules for deriving graphs. However, handcrafting such rules can become a tremendous effort — a well-known problem in the field of expert systems, called the “knowledge engineering bottleneck”. In this paper we discuss approaches to the automatic induction of production rules from given design graphs using machine learning. Four approaches were compared with respect to an application in conceptual design, specifically functional decomposition. The parse/derive method is an original contribution of this paper.

Keywords. Design Automation, Machine Learning, Graph-Rewriting, Functional Decomposition, Rule Induction

DOI: 10.14733/cadconfP.2016.157-161