NATURAL LANGUAGE PROCESSING FOR CONSTRUCTION SITES MANAGEMENT

Abstract

Information flow management has historically been a problem in the construction industry. Fragmentation of processes within the sector, interdisciplinarity, and lack of unambiguous methods for passing information are some of the obstacles that make it difficult to have a nonsplintered flow of data. BIM approach is, if well adopted, a valuable support for optimal information flow management. Studies suggest that the use of Natural Language Processing (NLP) with vocal assistants can facilitate the access and understanding of the information to non-technical users. This research proposes the development of a system for querying BIM models through natural language and without the use of specific keywords. Input is provided by spoken language, while output can be either spoken or visual, provided a graphic device is connected to the virtual assistant. This eases information retrieval in environments where it is difficult to use hand-held computing devices such as a construction site.