Machine learning in the discipline of architecture: A review on the research trends between 2014 and 2020. (March 2023)

• Record Type: Journal Article
• Authors: Özerol, Gizem; Arslan Selçuk, Semra
• Is Part Of: International journal of architectural computing. Volume 21:Number 1(2023)
• Page Start: 23
• Page End: 41
• Publication Date: 2023-03
• Subjects: Artificial intelligence -- machine learning -- deep learning -- architectural research -- bibliometric analysis; Architecture -- Data processing -- Periodicals; Architecture -- Informatique -- Périodiques; Virtual reality in architecture -- Periodicals; Computer-aided design -- Periodicals; Architecture -- Data processing; Periodicals; 720.2840285536
• DOI: 10.1177/14780771221100102
• Languages: English
• ISSNs: 1478-0771

Abstract

Through the recent technological developments within the fourth industrial revolution, artificial intelligence (AI) studies have had a huge impact on various disciplines such as social sciences, information communication technologies (ICTs), architecture, engineering, and construction (AEC). Regarding decision-making and forecasting systems in particular, AI and machine learning (ML) technologies have provided an opportunity to improve the mutual relationships between machines and humans. When the connection between ML and architecture is considered, it is possible to claim that there is no parallel acceleration as in other disciplines. In this study, and considering the latest breakthroughs, we focus on revealing what ML and architecture have in common. Our focal point is to reveal common points by classifying and analyzing current literature through describing the potential of ML in architecture. Studies conducted using ML techniques and subsets of AI technologies were used in this paper, and the resulting data were interpreted using the bibliometric analysis method. In order to discuss the state-of-the-art research articles which have been published between 2014 and 2020, main subjects, subsets, and keywords were refined through the search engines. The statistical figures were demonstrated as huge datasets, and the results were clearly delineated through Sankey diagrams. Thanks to bibliometric analyses of the current literature of WOS (Web of Science), CUMINCAD (Cumulative Index about publications in Computer Aided Architectural Design supported by the sibling associations ACADIA, CAADRIA, eCAADe, SIGraDi, ASCAAD, and CAAD futures), predictable data have been presented allowing recommendations for possible future studies for researchers.