T/SAIAS 034-2025 Guidelines for the construction of corpus on high-end equipment
T/SAIAS 034-2025 Guidelines for the construction of corpus on high-end equipment
Basic Information
Scope
This document stipulates the general principles of high-end equipment corpus construction, corpus resource standards, corpus collection, corpus production, and corpus security. This document applies to high-end equipment manufacturing enterprises, scientific research institutes, technical service agencies, and other entities engaged in high-end equipment corpus research, development, maintenance, application, and evaluation. It covers the collection, processing, annotation, storage, and application of multi-source and heterogeneous high-end equipment corpora in core processes such as R&D design, production manufacturing, and operation and maintenance services. The corpus construction work in other industrial sub-fields can also be used for reference. For the first time, this guideline systematically standardizes the technical requirements and full-process management of corpus resource construction in the high-end equipment field. The document clarifies the classification system of high-end equipment corpora, covering three core scenarios: design and R&D, production manufacturing, and operation and maintenance services. It also stipulates the standards for the collection, processing, annotation, and storage of multi-modal corpora such as text, images, videos, timelines, and CAX (CAD/CAE/CAM). At the same time, the guideline innovatively proposes a method of constructing corpora combining real-world collection and synthetic generation, and establishes a quality evaluation system centered on accuracy, consistency, and applicability. The implementation of this document will provide key guidance for high-end equipment manufacturing enterprises and scientific research institutes in building high-quality, high-value, and safe and compliant AI corpora, and strongly support the in-depth application and innovative development of industrial large models and intelligent systems in the high-end equipment field.