시멘틱 지식그래프 구축 및 활용을 통한 BIM 모델 내 간섭 유형 및 심각도 분류 방안

Semantic Knowledge Graph Construction and Utilization for Clash Type and Severity Classification in Federated BIM Models

Clash detection in BIM is a critical process for identifying potential interferences during the design phase and preventing design errors. However, conventional clash detection approaches are limited to simple geometric overlap checks and fail to consider contextual information such as clash types and severity, reducing their effectiveness in supporting design adjustments. To address these limitations, this study proposes the ‘BIMClash’ framework, which leverages a semantic knowledge graph to represent clash information and classifies both clash types and severity levels using Cypher-based automated queries. A domain-specific ontology schema was developed, and object and relational data were extracted from IFC-based BIM models to construct the semantic graph in Neo4j. The framework incorporates spatial adjacency, attribute data, and predefined classification criteria to enable automated clash classification. Experimental results demonstrated that BIMClash achieved over 90% accuracy in clash type and severity classification compared to expert annotations, while reducing classification time by approximately 84% relative to manual processes. These findings highlight the practical value of the proposed framework in enhancing both the accuracy and efficiency of clash detection during the design stage.