Abstract
Advancements in digitization are fostering innovation within Industrie 4.0 (I4.0) ecosystems. The 2030 vision for I4.0 prioritizes sovereignty, sustainability, and interoperability to transform value chains into dynamic networks. While many physical and syntactical interoperability challenges have been addressed, there are still open questions regarding semantic interoperability. Natural Language Processing (NLP)-based ranking by semantic similarity offers a straightforward approach for matching semantically heterogeneous data but often lacks informative rationale and usability, especially when relevant matches are buried deep in the rankings. In this paper, we explore how we can further improve the ranking to obtain more precise semantical matches by leveraging Large Language Model (LLM), in particular Retrieval-Augmented Generation (RAG). Since comprehensibility is crucial for the end user when selecting the matches, we employ the LLM to additionally explain the rankings and enable follow-up queries. In a series of experiments, we show that this not only improves the usability, but also ensures that users can navigate complex semantic matches effectively.