Context is all you need - Enhancing contextual awareness on sustainability requirements in product development using natural language processing

For the successful implementation of sustainable product development, sustainability aspects must be considered beginning at an early stage. Limited available information demands assistance to support the product developer in identifying and relating relevant contextual information. This paper explores the extent to which contextual information can be expanded and characterized at an early stage to assess potential influences on sustainability. Known context entities are first embedded, then related by linguistically similar entities. Retrieval augmented generation (RAG) prompting is used to characterize and map the entities based on selective source clusters. The entities are then analyzed considering the impact on the Sustainable Development Goals (SDGs) in product life cycle phases to subsequently derive sustainability requirements. For validation, the approach is implemented in an exemplary development of bicycles. This approach helps to systematically expand the context and to assess the impact of the contextual entities on sustainability based on relevant information. As a result, the situational awareness of product developers is enhanced, which supports the early derivation of sustainability requirements. Thus, the number of iteration cycles can be reduced.