Knowledge graph-based intelligent QA system for pavement design

Abstract Pavement design codes serve as critical references in road engineering, playing a significant role in ensuring design safety and project quality. However, the diverse and complex requirements of pavement design result in low query efficiency and the lack of an effective knowledge-sharing mechanism during code retrieval. Traditional textual storage of pavement design information hinders information correlation and efficient extraction. As a structured database, knowledge graph (KG) offers flexible storage, easy content expansion, and the ability to represent complex relationships, making them widely applicable in vertical domains. This study innovatively integrates Natural Language Processing (NLP) techniques for semantic extraction from asphalt pavement design codes and leverages the Neo4j graph database for structured storage and query answering—an approach with broad cross-domain application potential. Specifically, we convert textual content from asphalt pavement design codes into structured triples, visualize the KG via Neo4j, and develop an intelligent question-answering (QA) system. The system efficiently extracts key entities and inter-category relationships using NLP, enabling direct access to professional knowledge bases for precise query responses. This integration not only addresses the knowledge-sharing gap in traditional design practices but also promotes the intelligent development of pavement engineering, demonstrating scalability to other engineering domains.