Abstract

This topic is currently available as a request for the development of a DIN SPEC in accordance with the PAS procedure. The request is currently being reviewed. DIN SPECs are developed by DIN SPEC consortia (temporary committees). The document will be drafted and approved by the authors who adopt the business plan. This DIN SPEC is intended to define requirements for the use of large language models (LLMs) for the construction or validation of knowledge graphs (KGs) and - vice versa - for the use of knowledge graphs for generative large language models, e.g. to avoid hallucinations and generate factually more correct answers. Since this is a very dynamic and very active field of research, the planned DIN SPEC clarifies in particular the terminology, the possible applications as well as the basic application possibilities (LLMs for KGs and KGs for LLMs) and their respective advantages, disadvantages and challenges. The document is applicable to research, development and deployment of ML techniques, including but not limited to natural language processing (NLP), knowledge management and domain-specific applications using LLMs and KGs in the above sense. Occupational health and safety measures are not the subject of this document. This document does not specify any requirements for the safety (in the sense of safety and security) and health protection of employees. Furthermore, no safety levels and safety functions are defined for products or applications that contribute to the prevention of accidents or health hazards in the work context.

Begin

2024-09-27

Planned document number

DIN SPEC 91526

Project number

62022020