Integrating IFC and NLP for automating change request validations

Abstract

The management and the identification of design changes constitute an essential part of the of a design flow within the architecture, engineering and construction (AEC) industry, requiring the formalisation of a multi-disciplinary collaborative information modelling environment. Construction projects generate substantial amount of change information, which needs to be updated continuously throughout the process, from initial feasibility study to the decommissioning of facilities. Complications arise from the information storage in multiple incompatible file formats that can lead to the loss or omission of details. In addition to any unexpected changes, the mismanagement of information is another factor leading to delays and costly errors. In order to mitigate such issues, this paper proposes to integrate the Industry Foundation Classes (IFC) data model and Natural Language Processing (NLP) to validate and visually identify the result of change requests. The system is developed using C# by 1) integrating angular framework with ASP.NET (Active Server Pages) to create a dynamic single web page and 2). Using X-BIM toolkit that supports IFC format to read, create and visualise the BuildingSmart Data Models (aka IFC Models). This approach enables a web-based platform capable of generating reports and visual previews, highlighting the differences between IFC files throughout the design processes. A web-based system prototype allows users to compare subsequent versions of IFC design models in terms of additions, modifications and deletions. The prototype uses NLP to intelligently identify the changes that have been made as it compares newer and older versions of the same model, making this information available to designers and 3D modellers. Prospective work will focus on the application of artificial intelligence (AI) to automate the implementation of changes within the construction models.