Details of Research Outputs

For project stakeholders to do their engineering and management tasks, they need to be able to quickly find the building information they need. The emerging artificial intelligence (AI)-empowered natural language interface (NLI) has a significant potential to provide a time- and cost-efficient way to extract partial subsets of Building Information Models (BIM). However, although the existing studies have demonstrated different methods to retrieve BIM models via natural language queries (NLQs), what aspects of model information and constraints are valuable and required to be retrieved by end-users using natural language (NL) has not been explored. To address this gap, this study conducts an information requirement analysis to understand the information needs of NLIs in querying BIM models. This research starts with a survey of the existing literature on BIM-oriented query languages and query systems. 102 sample query sentences from these studies are then extracted and analyzed. Next, the information entities, constraints, and question types in the collected sentences are manually identified, classified, and counted. The results reveal which aspects of BIM model subsets, constraints, and question forms are frequently mentioned in common BIM queries, which provide crucial guidance for the future development of AI-driven NLIs for information support in BIM-based construction projects.