Advancing engineering research through context-aware and knowledge graph-based retrieval-augmented generation

Large language models (LLMs) are powerful in language understanding and content generation but frequently fall short of technical accuracy when they are applied to engineering code, standards, and design documents. To mitigate this, we are seeing the emergence of Retrieval-Augmented Generation (RAG) models that ground outputs of LLMs with information from external trustworthy resources, increasing the factual consistency. However, traditional RAG techniques are limited in the treatment of isolated information (limited to the amount of information in a fixed-size chunk) and are deemed ill-equipped to traverse semantically linked technical information. This study introduces a collection of new and highly deployable RAG-LLMs built on the n8n automation system and specifically designed for engineering domains. Framework effectiveness was tested on a set of prompts developed with the help of practicing electrical engineering professionals and should be read through the framework's lens for interpretation of national engineering codes, technical standards, and design standards. To mitigate the shortcomings of the conventional retrieval-based chunking methods, a contextual RAG-based approach is employed to align the retrieved content with the query context to improve relevance. Moreover, RAG is adopted to structure knowledge graph retrieval, which can retrieve densely linked concepts from multiple knowledge graphs, thereby promoting more profound semantic understanding in complex technical domains. The study describes the relative benefits of these improvements, points to practical deployment issues, strengths, and weaknesses. All the n8n workflows employed in this study are made available as supplementary materials to facilitate reproducibility and sharing within the engineering research community and practitioners.