Automating Design Review with Artificial Intelligence and BIM: State of the Art and Research Framework

摘要

Today's model checking applications are capable of checking building models for conformance to code clauses which restrict explicit dimension values, but very little more than that. Complex, implicit code clauses are still beyond the scope of such applications. Much research and development is needed for semantic enrichment (to pre-process models for checking) and for automating checking of complex conditions. A review of the state-of-the-art reveals that research and development has been limited to the application of symbolic AI methods, such as rule inferencing. Based on this review and on a set of early experiments designed to test basic feasibility, the authors propose a framework for research and development that specifically extends current thinking to include machine learning methods for both semantic enrichment and model review. Machine learning is shown to be applicable in situations where expression of the constraints in the form of explicit rules is impractical.