A multi-association rule-based knowledge graph method for exploring complex research trends in BIM

Abstract

Traditional scientometric analysis methods primarily rely on binary keyword associations, which limits the ability to capture complex research structures. This study proposes a multi-association rule-based method (MAR) that enables the extraction of higher-order (n-ary) associations among keywords to address the stated limitation. The method integrates the Apriori algorithm with knowledge graph visualization to quantitatively analyse multi-keyword associations. The proposed method is applied to the Building Information Modelling (BIM) research domain. The bibliometric data, comprising 12,185 publications, is retrieved from the Scopus database following a systematic refinement of the initial search results to the architecture, engineering, construction, and operation (AECO)-relevant literature. The results demonstrate that the MAR method effectively identifies complex research patterns. The framework advances traditional scientometric tools by enabling the identification of multi-keyword semantic associations with quantified strength metrics (support, confidence, lift). Emerging trends revealed by the regression analysis include the convergence of BIM with generative artificial intelligence, real-time data integration, explainable AI, and human-centric approaches. The findings provide a scalable and analytical approach for identifying interdisciplinary research trends and gaps. The proposed method enables a deeper understanding of complex knowledge structures while supporting strategic decision-making and technology adoption.