Automatic information retrieval with semantic analysis for green building evaluation

In the pursuit of sustainable cities, many countries have established the rating systems to label qualified projects as “green buildings” to promote the design and operation of sustainable building. However, the rating and labeling process is often time-consuming and skill-demanding due to the standards in multiple levels and the diversity of construction projects. This paper developed a framework that partially automated the process of green building evaluation to reduce repetitive and demanding manual work for rating experts. It used Natural Language Processing (NLP) to extract and organize semantic information from standards, and automatically identify and classify items. Computer algorithms and BIM technology could achieve the information of calculation reports and related drawings files extraction and evaluation. The proposed rulebased framework was verified by an example that automatically calculated the score in material saving. Compared with the results of manual review, each project could save an average of 110 minutes, and more than 80% of the projects could reach 90% accuracy. The framework can support evaluation for experts and promote the development of automated evaluation of green buildings.