Framework for Automatic Speech Recognition-Based Building Information Retrieval from BIM Software: Fid-Bau Portal

Framework for Automatic Speech Recognition-Based Building Information Retrieval from BIM Software

Shin, Sangyun / Lee, Chankyu / Issa, Raja R. A.

Since the appearance of automatic speech recognition-based virtual assistants (e.g., Google Assistant and iOS Siri), it has become more common to use voice-based information retrieval systems in search engines. Recently, such search engines have served not only to accept keyword-based commands from the human voice but also have evolved to recognize and understand more natural human language. In addition, various industries have tried to apply automatic speech recognition (ASR) systems to their fields to improve work efficiency or productivity. However, in the AECO field, there have not been many studies related to these systems. Most research in this field has been focused on using keyword-based commands rather than using natural human language for operating BIM S/W (e.g., Revit). This means that AECO research lag behind that of other fields trying to apply natural language-based ASR systems. Therefore, this paper proposes a method of operating BIM S/W using natural human language. It also is aimed at not merely accepting natural language but also understanding the intentions embedded in that natural language. For this purpose, this paper will use semantic-based BIM, which is implemented through the semantic parsing as a way of understanding intentions in recognized natural language. It will enable users to query a building information model more intuitively and directly find the building information they seek.