Ontology-based automated information extraction from building energy conservation codes

An ontology-based information extraction algorithm for automatically extracting energy requirements from energy conservation codes is proposed. The proposed algorithm aims to support fully-automated energy compliance checking in the construction domain by allowing automated extraction of the requirements from the codes instead of the status quo which relies on manual extraction of requirements from codes and manual formalization of those requirements in a computer-processable format. Automated information extraction from energy conservation codes, compared to other building codes, is a far complex task because many code provisions are long, hierarchically-complex, and with exceptions. A combination of text classification methods, domain-specific preprocessing techniques, ontology-based pattern-matching extraction techniques, sequential dependency-based extraction methods, and cascaded extraction methods is proposed to deal with such complexity in extraction. The proposed algorithm was tested in extracting energy requirements from Chapter 4 of the 2012 International Energy Conservation Code, and the results showed 97.4% recall and 98.5% precision.