Graph-based method for extracting spatial information from semi-formal text to derive 3D bridge and damage models from legacy maintenance data
Automation in Construction, vol. 178, pp. 106418
Abstract
Aging bridges require improved maintenance strategies; however, recent developments often rely on newly collected data to represent the bridge and its condition, hindering their large-scale adoption and thus significant improvements. This paper demonstrates how existing data from legacy bridge management systems (BMS) can be utilized to automatically create object-oriented knowledge graphs and three-dimensional models of bridge structures and their inspection data. It applies a relative spatial reference system to position and link components and damage, generating a bridge maintenance graph from BMS data that supports spatial queries using natural-language-based location terms. This enables the automatic localization of recorded damage through their textual location descriptions. The method successfully processed 90% of 2,348 damages from two use cases with a precision of 0.8 and a recall of 0.97. The approach bridges the gap between the needs of modern information models and legacy data structures, facilitating the widespread implementation of improved maintenance strategies. • Automatic creation of bridge models based on existing Bridge Management System data. • Automatic localization of damage based on natural language location descriptions. • Creates spatial knowledge graphs from bridge maintenance data. • Transforms legacy domain database structures into object-oriented models. • Integration of historical data in modern information management.
Authors 2
-
Affiliation as printed
Chair of Design Computation, RWTH Aachen University, Schinkelstraße 1, Aachen, 52062, Germany
-
Affiliation as printed
Chair of Design Computation, RWTH Aachen University, Schinkelstraße 1, Aachen, 52062, Germany
Cited by 6 stored of 6
6 results
No patents citing this paper on Lens.org (checked 2026-10-06).