Work

Ontology-Driven Mixture-of-Domain Documentation: A Backbone Approach Enabling Question Answering for Additive Construction

Year: 2025

Type: article

Abstract: Advanced construction techniques, such as additive manufacturing (AM) and modular construction, offer promising solutions to address labor shortages, reduce CO2 emissions, and enhance material efficiency. Despite their potential, the adoption of these innovative methods is hindered by the construction industry’s fragmented expertise. Building Information Modeling (BIM) is frequently suggested to integrate this diverse knowledge, but existing BIM-based approaches lack a robust framework for systematically documenting and retrieving the cross-domain knowledge essential for construction projects. To bridge this gap, this paper presents an ontology-driven methodology for documenting and utilizing expert knowledge, with a focus on AM in construction. Based on a well-founded ontological framework, a set of modular ontologies is formalized for individual domains. Additionally, a prototypical documentation tool is developed to elevate recorded information and BIM models as a knowledge graph. This knowledge graph will interface with advanced large language models (LLMs), enabling effective question answering and knowledge retrieval.

Source: Buildings

DOI: 10.3390/buildings15010133

Authors: Chao Li, Frank Petzold

Institution: Technical University of Munich

Language: English

Topic: BIM and Construction Integration

Subfield: Building and Construction

Field: Engineering

Domain: Physical Sciences

SDG: Industry, innovation and infrastructure

Open Access status: gold

Estimated APC paid: $3,137