|Ph.D Student||Michael Belsky|
|Subject||A Framework for Semantic Enrichment of IFC Building Models|
|Department||Department of Civil and Environmental Engineering||Supervisor||Professor Sacks Rafael|
|Full Thesis text|
The Architecture, Engineering and Construction (AEC) industry consists of separate parties that work together to bring a project to fruition. As a result smooth and reliable information exchanges between different parties involved in construction projects become vital. The primary software tools for design and construction are Building Information Modeling (BIM) tools, which are used by different parties involved in construction projects. The information in a project is generated in a wide variety of domain-specific software tools used by separate project stakeholders, and this information must be exchanged among them. These software tools each create and store the information in their native formats, which imposes serious challenges for information exchanges.
The leading AEC industry exchange standard (ISO 19739) for use with BIM models, the Industry Foundation Classes (IFC), is too broadly generic to capture the full semantic meaning of exchanged models in a strict way. Although BIM standards that prescribe Model View Definitions (MVD) for domain-specific exchanges in a precise way have been developed and/or are under development, insufficient interoperability remains a serious obstacle for achieving the full potential of BIM almost twenty years since the IFC schema was conceived.
This research proposes an innovative geometry and spatial topology driven approach for supplementing an IFC exchange file with semantically useful concepts defined for a receiving application and inferred from the explicit and implicit information contained in a building model exchange. This places the onus for interpreting data exchanges on the receiving application instead of the exporting application, which aligns better with the economic interests of BIM software companies. A prototype software was implemented to test the applicability of the approach. Among others, it consists of a rule-processing engine and allows compilation of inference rule sets that can be tailored for different domains. The rule sets encapsulate the knowledge of experts in a given domain, who are skilled at identifying the semantics of building objects in the context of building model. An inference rule is compiled in the form of IF-THEN statements in a language close to plain English. The acronym for the implemented software, ‘SEEBIM’ (Semantic Enrichment Engine for Building Information Modeling), reflects the notion that domain experts (architects, engineers, etc.) can ‘see’ information in a model that is implicit as well as the information that is carried explicitly.
The thesis formulates the approach for semantic enrichment of IFC models, describes the implementation of SEEBIM and illustrates the approach with examples of information exchanges from different domains. Samples of knowledge from these domains were encapsulated in the form of rule sets. The rule sets were executed by SEEBIM on a number of IFC files. Experiments to date showed that semantic enrichment of IFC models can be done automatically once inference rule sets are established. The library of topological, geometric and other generic operators, which are used for rule set compilation, can be reused across different domains. This suggests that although small in number, the set of operators is sufficiently generic to allow broad application across multiple domains in the AEC industry.