Automated Extraction of System Structure Knowledge from Text

AbstractThis paper presents a method to automatically extract structure knowledge of mechanical systems from natural language text. The current work extends our prior work on extracting function knowledge from text, which was presented at last year’s conference. The method uses rules based on a combination of syntactic, lexical, and redundancy information to identify structure knowledge from parsed text. Three case studies were conducted to evaluate the method. The case studies involved extracting physical connections among a known set of components of a bicycle frame, an internal combustion engine, and a drum brake from Wikipedia. The current work makes progress toward addressing the challenge of knowledge acquisition for knowledge-based CAD systems.

Download publication

Get in touch

Something pique your interest? Get in touch if you’d like to learn more about Autodesk Research, our projects, people, and potential collaboration opportunities.

Contact us