Automatic Information Extraction from Construction Quality Inspection Regulations: A Knowledge Pattern–Based Ontological Method

Abstract

Quality compliance checking is essential to ensure construction quality, the prerequisite for which is information extraction from construction quality inspection regulations (CQIRs). Due to the inclusion of multiple qualitative constraints, complex syntax, semantic structures, and exceptions, extracting constraint information from CQIR automatically is difficult. To address the research gap, a knowledge pattern–based ontological method was developed to extract constraint information automatically from CQIR. The entire study process was guided by design science. To begin, knowledge patterns of three typical types of construction quality constraints were investigated to identify constraint elements and their semantic relationships, namely construction procedure constraints, product quality attribute constraints, and resource selection constraints. Then an ontology model was developed to represent these knowledge patterns by defining concepts and properties based on identified constraint elements and semantic relations. Based on the proposed ontology model, Java Annotation Patterns Engine (JAPE) rules were encoded to extract constraint information from CQIR. Finally, a prototype system was created to validate the proposed method, using text data from five mandatory regulations of groundwork and foundation construction. Experimental results demonstrated the theoretical feasibility of the presented method in automatically extracting constraints from CQIR.