ABSTRACT
Assessing project conformity involves extracting information from building regulations and creating rules to verify it. Nowadays, automated systems are already available that assess project conformity but require manual intervention on rule creation, making the process time-consuming and prone to errors. To solve this limitation, this research proposes a new system for extracting information from regulatory codes, which combines the OpenIE6 model with rule-based NLP methods. The articles considered for the search are those containing quantitative values. They were used to train the proposed models to ‘learn’ the context of words in a sentence or document. From the normative articles, triplets (subject-relation-object) were automatically extrapolated and used for the creation of conformity rules. A case study is proposed in which the new data mining technology is applied and the conformity analysis is performed. The research is part of a larger project that aims to make the entire compliance process automatic.
