A FRAMEWORK FOR SEMANTIC ENRICHMENT OF IFC BUILDING MODELS

Building information models provide semantically rich information with objects that explicitly represent 3D geometry and non-graphical properties. However, there are significant difficulties in exchanging information between domain-specific BIM tools. Exchange formats, such as the Industry Foundation Classes (IFC) Coordination View (CV), cannot represent all of the domain- and profession-specific information required by the different construction project stakeholders (and their diverse software tools). Similarly, building models compiled from laser scanning point clouds or other survey methods often contain the major building objects but lack the semantic richness of relationships and other concepts needed for a model to be useful for a range of engineering purposes. All these limit the ability of diverse systems to interoperate together. An automated or semi-automated approach is therefore needed for deriving a semantically useful building model file from the explicit and implicit information contained in an IFC building model. In this paper, we present such an approach for semantic enrichment of IFC building models exported according to the IFC CV version 2. The innovation of the approach is that it supplements the exchange model data with semantic constructs defined for the receiving application. This places the emphasis for data exchange on the receiving application by moving the onus from export to import. The approach is based on a taxonomy of semantic inference rules. A three-tier structure for the taxonomy is proposed in order to capture the domain information in a structured way. The paper outlines the approach and illustrates it with examples of exchanges from the precast concrete domain defined by the US National BIM Standard (NBIMS). The digital information content needed to support these exchanges is specified in Model View Definitions (MVDs). Intermediate results of a pilot technical implementation for the examples are presented and analyzed in the paper.