The X-Formats Family: Linking Geometry and Semantics for Product-Centric Engineering Data Management

Machine-readable semantic information, such as simulation settings, inspection configurations, machine parameters, process outcomes/logs, is central to digital manufacturing, enabling automation, interoperability, data traceability, and adaptive reconfiguration across the product lifecycle. When anchored to specific geometric entities, these semantic structures form the backbone of Model-Based Definition (MBD), a data engineering architecture in which the CAD model acts as the single source of truth to realize the digital thread. Within this architecture, both rule-based and machine-learning pipelines can orchestrate simulation, optimization, and control tasks using shared, semantically enriched models. Despite this potential, most engineering file formats either lack expressive, machine-readable semantics or encode them in ways that are format-specific, difficult to adapt to different needs, and hard to reuse. This letter presents a manifesto for the eXtended Formats (X-Formats) family, a lightweight stand-off markup approach that links structured annotations to engineering files without modifying the original source. Semantics can be defined independently of the underlying syntax (e.g., STL, B-rep, STEP, and DXF for CAD; PNG, JPEG, GIF, and SVG for images; MP4 and MOV for videos; GeoTIFF for geospatial data) and are anchored to file-specific entities that can be detected by external feature-extraction mechanisms. X-Formats structure semantics in a three-level hierarchy composed of property, layer, and schema, which is fundamental for supporting modular dataset design and allowing rapid customization across domains by replacing only the geometry-descriptor block. This letter introduces the core architectural principles, anchoring strategy, supporting tooling, and broader implications of this novel, product-centric, approach to engineering data management.