Text-based automatic knowledge graph construction for road infrastructure operations management

Automatic construction of knowledge graphs (ACKG) from text enables intelligent operations management of road infrastructure (OMRI). The specialized nature of OMRI text hinders direct adoption of general ACKG methods and necessitates domain-specific approaches. The rapid evolution of OMRI-specific ACKG renders a review necessary. This paper aims to summarize the latest progress and to guide future ACKG research for OMRI applications. 41 articles from seven databases (2020–August 2025) are analyzed systematically. The review provides an in-depth analysis of design motivations and underlying mechanisms of the methods involved, maps the approaches to challenges from textual characteristics, and proposes a domain-tailored process architecture. Key findings include: (1) adoption of advanced technologies, particularly machine learning, addresses domain challenges and facilitates automation; (2) the “extraction-generation-refinement” workflow forms a reusable roadmap; (3) four key aspects reflect ACKG methods' effectiveness; and (4) remaining challenges include technology coverage, and promising directions include transfer learning. • A systematic review on ACKG from texts in OMRI is conducted. • ACKG methods are analyzed based on characteristics of OMRI texts. • A three-phase ACKG process architecture for OMRI texts is proposed. • Four aspects of ACKG effectiveness evaluation in OMRI scenarios are identified. • Five challenges and ten future directions for ACKG on OMRI texts are outlined.