FROM TEXT TO IMAGES: LINKING SYSTEM REQUIREMENTS TO IMAGES USING JOINT EMBEDDING

Abstract Smart manufacturing enterprises rely on adapting to rapid engineering changes while minimizing the generated risk. Making informed decisions related and managing risks against unexpected costs requires more information be extracted from limited data. However, in early-stage design can come many forms, namely text images. The development of innovative tools processes link multisource data together is essential assist designers building model-based (MBE) systems. formal computational linking yet realized MBE. We propose a framework implement transfer learning integrate domain specific knowledge bridge this gap. A synthetic dataset created using web scraping techniques based keywords requirements. Requirement-image pairs are used fine tune contrastive language-image pretraining model acquire knowledge. results demonstrate how content images indicate all affected requirements for tracing complex system.