Files
Abstract
In engineering design, standards provide essential technical definitions and guidelines, reflecting best practices recognized across the industry. These standards, developed by various Standard Developing Organizations (SDOs), encapsulate technological expertise to enhance safety, reliability, productivity, and efficiency in component and system design. However, the vast number and diversity of these documents pose challenges for designers in selecting and implementing appropriate standards due to inconsistencies, conflicting advice, and overlaps. This thesis proposes an information retrieval approach to improve product development by linking product requirements with relevant engineering standards. It explores the use of Artificial Intelligence techniques, including language embeddings, semantic search, and zero-shot learning, to efficiently navigate this extensive information corpus. Additionally, the study investigates the optimal contextual window using sentence-level and extensive project data, aiming to develop novel frameworks for more effective engagement with the textual knowledge in engineering standards, thereby aiding design engineers.}%