This paper addresses the critical challenge of data validation in the Architecture, Engineering, and Construction (AEC) industry, arising from the interoperability issues linked with Building Information Modeling (BIM) and Industry Foundation Classes (IFC) standards. Despite the potential of IFC in improving project lifecycle management, the accuracy and reliability of BIM data remain hindered by insufficient validation tools. The paper proposes a novel approach employing deep graph convolutional neural networks (DGCNNs) to validate and correct IFC model properties, leveraging 3D model element segmentation. This methodology aims to enhance data reliability, facilitating improved interoperability within the AEC sector. By examining the feasibility of neural networks for property validation and by converting neural network results into actionable model element properties, the research contributes to advancing BIM environments towards greater accuracy and efficiency. The implications of this study extend to improving project delivery times, reducing costs, and enhancing collaboration among stakeholders.