Information requirement extraction out of regulatory documents with text-based artificial intelligence

Information requirements and regulations for buildings are formulated in building codes and standards. They are used, among others, during the building permit or Building Information Modeling (BIM) processes to verify information requirements and technical specifications.

Nevertheless, information requirements are mostly formulated in natural language and non-machine-readable way. Because of that, the information needs to be extracted manually for usage in the desired processes by experts in a laborious, and error-prone way.

Therefore, my research aims to create a Natural Language Processing (NLP) pipeline for the extraction of the information requirements out of the building codes and standards written in German. In the next step, the information is converted into an interchangeable, machine-interpretable, and effectively usable format as a semantic knowledge base for the usage in different BIM-processes.