Information Extraction for Knowledge Graph of ISO 19650 Standards

Wu Bing,Song Yuanbin,Chu Junyi,Cao Jinhao

DOI: https://doi.org/10.1007/978-981-19-3486-5_20

2022-01-01

Abstract:Engineering project development requires all participants timely communicate explicit information. Besides the IFC file format, they also need an information management framework to support their collaboration. ISO 19650 standard series provide such a framework to establish a reliable information source. Since the 5-part ISO 19650 series constitute a complex system, the AEC community expect a knowledge graph to capture the key concepts and rules in those standards in order to facilitate the application of the collaborative management framework. Unfortunately, manual development of ISO 19650 knowledge graph is costly and time consuming. Therefore, a NLP-based information extraction method has been specifically developed for automatic construction of the ISO 19650 knowledge graph. In particular, the difficulty arising from the lack of corpus is greatly overcome by reasoning out domain semantic relationships from the syntactic relationships with the mapping rules specifically developed in this study. Finally, the experiment verified the feasibility and usefulness of the developed information extraction method.