Automated generation of building fingerprints using a spatio-semantic query language for building information models

In early stages of building design, architects frequently make use of referential solutions as a source of inspiration. Today, these reference solutions are looked up manually in a time-consuming and laborious process. To speed up the process, means for automating this procedure are required. The approach discussed here is based on the assumption that the reference solutions (i.e. building designs) are stored in a repository as Building Information Models (BIM) using the open data model Industry Foundation Classes (IFC). In order to find suitable reference solution for a given problem, a measure of similarity has to be defined. To this end, we introduce the notion of the building fingerprint as a way to capture the main characteristics of a building design. A major component of the fingerprint is the representation of the adjacency relationships between the spaces, which can be expressed by an adjacency graph. Another component is the accessibility relationships, which again can be expressed by a corresponding graph. To use this graphs as a basis for the lookup mechanisms, they have to be generated in an automated way for the large set of IFC models stored in the reference solution repository. In this paper, we present an approach based on the BIM query language QL4BIM which provides high-level topological operators to efficiently retrieve accessibility and adjacency relationships among spaces. We discuss how QL4BIM is applied to generate the graph components of the building fingerprints.