Automation in Construction (2000–2023): Science Mapping and Visualization of Journal Publications

Abstract

This paper presents a scientometric review that provides a quantitative perspective on the evolution of Automation in Construction Journal (AICJ) research, emphasizing its developmental paths and emerging trends. The study aims to analyze the journal’s growth and citation impact over time. It also seeks to identify the most influential publications and the cooperation patterns among key contributors. Furthermore, the study explores the journal’s primary research themes and their evolution. Accordingly, 4084 articles were identified using the Web of Science (WoS) database and subjected to a multistep analysis using VOsviewer version 1.6.18 and Biblioshiny as software tools. First, the growth and citation of the publications over time are inspected and evaluated, in addition to ranking the most influential documents. Second, the co-authorship analysis method is applied to visualize the cooperation patterns between countries, organizations, and authors. Finally, the publications are analyzed using keyword co-occurrence and keyword thematic evolution analyses, revealing five major research clusters: (i) foundational optimization, (ii) deep learning and computer vision, (iii) building information modeling, (iv) 3D printing and robotics, and (v) machine learning. Additionally, the analysis reveals significant growth in publications (54.5%) and citations (78.0%) from 2018 to 2023, indicating the journal’s increasing global influence. This period also highlights the accelerated adoption of digitalization (e.g., BIM, computational design), increased integration of AI and machine learning for automation and predictive analytics, and rapid growth of robotics and 3D printing, driving sustainable and innovative construction practices. The paper’s findings can help readers and researchers gain a thorough understanding of the AICJ’s published work, aid research groups in planning and optimizing their research efforts, and inform editorial boards on the most promising areas in the existing body of knowledge for further investigation and development.

1. Introduction

Automation in Construction (AICJ) is a leading international journal with an established position in international listings, including the Clarivate Journal Citation Report (JCR) and Scopus Source CiteScore report (SCR). AICJ is recognized among the first quartile sources in the “construction building technology” (top 5) and “civil engineering” (top 5) categories in Clarivate JCR where it emerges among the first quartile sources in the “civil and structural engineering” (top 5), “building and construction” (top 5), and “control and system engineering” (top 15) categories in Scopus SCR. AICJ provides peer-reviewed content on the application of information and communication technologies (ICT) in design engineering, construction, and the maintenance and management of constructed facilities. AICJ’s publication encompasses all phases of the construction life cycle, starting with conceptual planning and design, through construction and execution, to the operation and maintenance of the facility, and finally, deconstruction and recycling of the building structures. This scope has led to a vast number of publications (4084 articles) over the last two decades with high practical relevance in diverse research domains including but not limited to computer-aided design, decision support systems, classification and standardization, simulation models, robotics, supply chain management, automated inspection, facilities management, information systems, and intelligent systems. Despite being one of the most reliable sources of knowledge and innovation in the construction and building field over the past two decades, there is a lack of studies that provide a comprehensive quantitative analysis of AICJ literature. To address this gap, the current research offers a novel scientometric analysis of the AICJ research studies, focusing on the following research questions:

• How has AICJ’s research productivity and influence evolved from 2000 to 2023?
• Which are the influential AICJ articles?
• What are the pioneering research entities published in AICJ in terms of counters, organizations, and authors?
• What is the conceptual knowledge structure of AICJ articles, and how has it dynamically evolved from 2000 to 2023?

Scientometric analysis, as a subfield of information science, employs quantitative methods and bibliometric techniques to examine and visually depict significant patterns, emerging trends, and the evolution of knowledge structures associated with a specific research entity by analyzing extensive bibliographic records [1,2,3]. This analysis is typically performed on the titles, keywords, abstracts, and/or citation records of bibliographic records, as these aspects are considered to provide an obvious and succinct description of the publications’ substance and direction [4,5,6]. The terms “scientometric review” and “bibliometric review” are often used interchangeably to describe the quantitative analysis of scientific literature. In contrast, “science mapping” specifically refers to the visual representation of research structures, thematic patterns, and intellectual linkages over time.

In academia, it is a prevalent practice to conduct scientometric reviews or analyses for journals with a recognized academic history, serving as a quantitative appraisal of previous research efforts that guide future directions. For instance, Donthu et al. [7] reviewed the Journal of Business Research, using 5344 documents published between 1972 and 2017. Sigala et al. [8] presented a bibliometric overview of the Journal of Hospitality and Tourism Management, utilizing 537 papers published between 2006 and 2020. Xue et al. [9] conducted a bibliometric analysis of the principal international journal Process Safety and Environmental Protection, using 3152 documents from 1990 to 2020. Donthu et al. [10] conducted a bibliometric analysis of the International Journal of Information Management over a 40-year period, spanning from 1979 to 2019, based on 1629 publications. El-adaway et al. [11] reviewed the publications of the Construction Management and Economics Journal using social network analysis (SNA), descriptive analysis, and predictive machine learning (ML) over 40 years, encompassing 2207 publications.

The similarities between the studies above lie in their reliance on reliable data sources, the use of large bibliographic datasets spanning extended time horizons, the utilization of open or closed bibliometric tools (VOSViewer, Biblioshiny, and CiteSpace), and a heavy reliance on statistical measures. However, these studies often provide a limited and confined in-depth exploration in terms of focal research topics and concepts. Building on these points and aligned with the four established research questions, this study conducts a comprehensive analysis and evaluation of (1) the publication and citation trends of AICJ over time; (2) the most influential AICJ documents, identified through global citation score (GCS) and document average citations per year (DACY); (3) the scientific collaboration networks among countries, institutions, and authors using co-authorship analysis; and (4) the thematic structure of AICJ publications, explored through author-keyword (AK) co-occurrence and thematic evolution analyses. The findings of this study offer valuable insights by helping researchers better understand the development of AICJ’s scholarly output, supporting research groups in strategic planning, and guiding editorial boards to prioritize emerging and impactful research directions within the journal’s scope. It is worth mentioning that the focus of the study is on AICJ due to its recognized leadership and substantial influence in the construction research field, as well as its ranking within the top quartile categories in JCR and SCR. Moreover, AICJ significantly shapes and represents evolving research directions in construction research and technology, making it significant in conducting focused, in-depth scientometric analysis. The remaining sections of this paper are structured in the following manner. Section 2 describes the research methodology, data acquisition procedures, and the methods and software tools employed. Section 3 encompasses the results and findings derived from the scientometric analysis. Section 4 and Section 5 present comprehensive discussions and conclusions.

2. Materials and Methods

2.1. Methodology Description

This study employs a three-stage approach: Stage I entails the acquisition of bibliographic data from the WoS Core Collection database. Then, Stage II encompasses appointing the analytical procedures and techniques. Stage III involves the evaluation of data and findings through these three steps: (I) exploring AICJ publication and citation over time and recognizing influential documents; (II) exploring the scientific cooperation networks regarding countries/regions, organizations, and authors; and (III) science mapping AICJ publications using author-keyword co-occurrence and thematic evolution analyses.

2.2. Data Collection

This study follows the PRISMA 2020 guidelines to conduct and report the scientometric review (see Figure 1). As the aim is to extract publications from the AICJ journal, the number of publications remains consistent across all bibliographic databases during the study period (2000–2023). Accordingly, the governing factor is the quality of extracted bibliographic data. The study relies on the WoS Core Collection database for bibliographic data acquisition. The rationale behind this is that WoS ensures the uniformity and reliability of bibliographic data with respect to authors (full names), organizations (unified names), and countries/regions, as guided by Visser et al. [12]. To retrieve the AICJ publications from the literature search engine, the search term “Automation in Construction” is used, along with the following search statement:

Figure 1. PRISMA studies’ inclusion sequence.

Figure 1. PRISMA studies’ inclusion sequence.

Using the aforementioned phrase, the literature search is conducted on the publication title (SO). The date range (DOP) is defined from 1 January 2000 to 31 December 2023. As for the document type (DT), only peer-reviewed journal articles are considered. The inclusion of various document types such as “review”, “letter”, and “editorial material” would compromise the consistency of the extracted records and result in misleading study findings. Furthermore, peer-reviewed articles reflect novelty and originality in research works. Therefore, a total of 4084 documents are defined and extracted. All bibliographic data for these documents, including full records and cited references, are extracted for conducting the analysis. The quality of bibliographic data—covering aspects such as Author, Document Type, Journal, Language, Publication Year, Scientific Categories, Title, Total Citations, Cited References, Abstract, DOI, Affiliation, Corresponding Author, and Keywords—are analyzed and reported using the Biblioshiny tool, as detailed in

Table 1. AICJ articles’ data quality report.

Description	Completeness	Status
Author	100.00%	Excellent
Document Type	100.00%	Excellent
Journal	100.00%	Excellent
Language	100.00%	Excellent
Publication Year	100.00%	Excellent
Science Categories	100.00%	Excellent
Title	100.00%	Excellent
Total Citation	100.00%	Excellent
Cited References	99.93%	Good
Abstract	99.90%	Good
DOI	99.88%	Good
Affiliation	99.63%	Good
Corresponding Author	99.56%	Good
Keywords	98.53%	Good

. The results indicate a high level of completeness, with percentages ranging from 98.53 to 100%.

2.3. Methods and Software Tools

The bibliometric methodologies employed in accordance with the research objectives include co-authorship analysis, keyword co-occurrence analysis, and thematic evolution of keywords.

• Co-authorship analysis constitutes a quantitative technique that examines the cooperation patterns among the authoring entities based on the number of common publications [13,14,15].
• Keyword co-occurrence analysis is a quantitative technique that quantifies the mutual occurrence of keywords by examining the number of common publications [14,16].
• Keywords thematic evolution is a quantitative technique that examines and investigates the development of keywords throughout a series of subsequent time-slices [17,18,19].

There are several trustworthy bibliometric software tools available, offering a range of features and advantages. This study visualizes and analyzes the 4084 AICJ papers identified using Biblioshiny and VOSviewer tools.

• VOSviewer is a software tool that can handle large amounts of data while producing and visualizing bibliometric networks in an approachable manner using the VOS mapping technique. Using the co-authorship analysis function, VOSviewer serves as a tool for mapping and visualizing the scientific collaboration among countries, organizations, and researchers. Additionally, it uses the co-occurrence analysis function to discover and visualize the links between author-keywords in publications.
• Biblioshiny is a web-based interface that employs the primary features of the bibliometrix R-package, which is provided by Aria and Cuccurullo [17], to conduct science mapping analysis. In this study, the Biblioshiny thematic evolution function is employed to investigate and illustrate the temporal progression and distribution of AICJ research themes based on author-keywords.

Table 2. VOSviewer–Biblioshiny comparison.

Aspect	VOSviewer		Biblioshiny
Applicability	Co-authorship analysis	✓	Co-authorship analysis	✓
	Co-occurrence analysis	✓	Co-occurrence analysis	✓
	Thematic evolution	-	Thematic evolution	✓
Strength	Open source Figures quality Results easy-extraction Results easy-interpretability Friendly user interface		Open source Results easy-extraction Friendly web-based interface
Limitation	Disallowance of naming the network’s clusters Inability to perform thematic evolution analysis		R language expertise

provides a detailed comparison between VOSviewer and Biblioshiny, explaining their applicability, strengths, and limitations. VOSviewer excels in generating high-quality figures, conducting co-authorship and co-occurrence analyses, but lacks thematic evolution capabilities and cluster naming. On the other hand, Biblioshiny supports thematic evolution and offers a user-friendly web-based interface, but it requires expertise in the R language. In this study, the authors leverage the strengths of both tools to match their methodological requirements and serve the research objectives.

3. Analysis and Findings

The Analysis and Findings section is organized into three major sections, each addressing a definite aspect of the study objectives. Section 3.1 focuses on publication and citation analysis, examining the growth trends in publications and citations while identifying influential studies that shape the research landscape. Section 3.2 delves into the co-authorship analysis, exploring key contributors at the levels of countries, organizations, and individual authors to highlight collaborative patterns. Finally, Section 3.3 presents a science mapping of AICJ publications, leveraging keyword co-occurrence analysis and keyword thematic evolution mapping to uncover trends and thematic shifts in research topics over time.

3.1. Publication and Citation Analysis

Annual growth of publications and citations is a significant indicator that reflects the knowledge maturity, accumulation, and of a specific research entity (e.g., country, organization, publisher, or source) [15,20,21]. In contrast, identifying influential articles is salutary for determining the key scientific areas for a research entity [9,22]. In this light, the Annual growth of publications and citations, and influential articles are analyzed and provided in the following sections.

3.1.1. Publication/Citations Growth

As depicted in Figure 2, from January 2000 to December 2005, the number of publications is 322 articles, constituting 7.9% of the total AICJ documents for the study period (2000–2023). The total citations are 538, representing less than 0.5% of the total AICJ citations. From January 2006 to December 2011, the number of publications increased to 536 articles, which constitute 13.1% of the overall AICJ documents. The total citations equal 5492, representing 3.9% of the total AICJ citations. Between January 2012 and December 2017, the number of publications nearly doubled compared to the prior period (2006–2011), with a total of 1001 published articles. This accounts for 24.5% of the AICJ documents during the study period from 2000 to 2023. In contrast, the total citations of this period are 25,257, representing 17.7% of total AICJ citations. The number of publications has increased by 2.22 times in the past five years (January 2018–December 2023) compared to the previous period (2012–2017), with a total of 2225 articles, corresponding to approximately 54.5% of the AICJ documents from the 2000–2023 study period. On the other hand, the total citations for this period are 4.4 times higher than those for the previous period (2012–2017), reaching 111,091 and representing 78.0% of the total AICJ citations for the study period (2000–2023). The rapid growth of publications and citations, particularly over the last decade, indicates that the AICJ journal has become increasingly popular and multidisciplinary, with a high global influence on the field of research in construction and building technology.

3.1.2. Influential Articles

Two main metrics, the global citation score (GCS) and the document average citations per year (DACY), are used to identify the most influential articles. The GCS is a direct indicator that grasps the overall scientific value and impact of a publication. Such an indicator requires time, supposedly years, to aggregate [23]. Consequently, the DACY is computed and used in conjunction with the GCS to address the publication year effect. According to the analyzed indicators,

Table 3. Top influential AICJ publications.

Title	Year	DOI	GCS	DACY
Autonomous concrete crack detection using deep fully convolutional neural network	2009	10.1016/j.autcon.2008.10.003	768	51
The gap between predicted and measured energy performance of buildings: A framework for investigation	2014	10.1016/j.autcon.2014.02.009	649	65
Developments in construction-scale additive manufacturing processes	2019	10.1016/j.autcon.2018.11.028	648	130
Computer vision-based concrete crack detection using U-net fully convolutional networks	2012	10.1016/j.autcon.2011.06.010	636	53
Building information modeling framework: A research and delivery foundation for industry stakeholders	2004	10.1016/j.autcon.2003.08.012	585	29
Mobile 3D mapping for surveying earthwork projects using an Unmanned Aerial Vehicle (UAV) system	2014	10.1016/j.autcon.2014.01.004	518	52
Blockchain-based framework for improving supply chain traceability and information sharing in precast construction	2013	10.1016/j.autcon.2012.05.006	488	44
Building information modeling (BIM) and Safety: Automatic Safety Checking of Construction Models and Schedules	2010	10.1016/j.autcon.2010.09.002	480	34
Detecting non-hardhat-use by a deep learning method from far-field surveillance videos	2013	10.1016/j.autcon.2012.10.006	432	39
Falls from heights: A computer vision-based approach for safety harness detection	2019	10.1016/j.autcon.2019.04.005	418	84
A deep hybrid learning model to detect unsafe behavior: Integrating convolution neural networks and long short-term memory	2013	10.1016/j.autcon.2013.09.001	395	36
Automatic creation of semantically rich 3D building models from laser scanner data	2011	10.1016/j.autcon.2010.09.011	380	29
A BIM-data mining integrated digital twin framework for advanced project management	2012	10.1016/j.autcon.2012.02.008	368	31
BIM implementation throughout the UK construction project lifecycle: An analysis	2011	10.1016/j.autcon.2010.09.019	348	27
Building information modeling (BIM) for green buildings: A critical review and future directions	2009	10.1016/j.autcon.2008.07.003	337	22
Building information modeling for sustainable design and LEED (R) rating analysis	2018	10.1016/j.autcon.2017.11.002	302	50
Understanding and facilitating BIM adoption in the AEC industry	2011	10.1016/j.autcon.2010.09.016	301	23
Automated construction by contour crafting-related robotics and information technologies	2018	10.1016/j.autcon.2017.09.018	300	50
Public and private blockchain in construction business process and information integration	2007	10.1016/j.autcon.2006.12.010	295	17
The value of integrating scan-to-BIM and scan-vs-BIM techniques for construction monitoring using laser scanning and BIM: The case of cylindrical MEP components	2015	10.1016/j.autcon.2014.05.014	293	33
Technology adoption in the BIM implementation for lean architectural practice	2020	10.1016/j.autcon.2019.103063	243	61
Hybrid pixel-level concrete crack segmentation and quantification across complex backgrounds using deep learning	2020	10.1016/j.autcon.2020.103291	206	52
Sustainable performance criteria for construction method selection in concrete buildings	2020	10.1016/j.autcon.2020.103254	205	51
Autonomous pro-active real-time construction worker and equipment operator proximity safety alert system	2020	10.1016/j.autcon.2020.103276	200	50
A UAV for bridge inspection: Visual servoing control law with orientation limits	2020	10.1016/j.autcon.2020.103087	196	49
Freeform construction: Mega-scale rapid manufacturing for construction	2021	10.1016/j.autcon.2021.103564	186	62
Integrated digital twin and blockchain framework to support accountable information sharing in construction projects	2021	10.1016/j.autcon.2021.103688	166	55
Automatic crack classification and segmentation on masonry surfaces using convolutional neural networks and transfer learning	2021	10.1016/j.autcon.2021.103606	151	50
Attention-based generative adversarial network with internal damage segmentation using thermography	2022	10.1016/j.autcon.2022.104412	96	48
Automatic recognition of pavement cracks from combined GPR B-scan and C-scan images using multiscale feature fusion deep neural networks	2023	10.1016/j.autcon.2022.104698	66	66
Automatic pixel-level detection of vertical cracks in asphalt pavement based on GPR investigation and improved mask R-CNN	2023	10.1016/j.autcon.2022.104689	52	52
Pavement crack detection based on transformer network	2023	10.1016/j.autcon.2022.104646	51	51

presents the top 20 publications, sorted by the highest GCS, and the top 20 publications, sorted by the highest DACY. Notably, 19 of the most widely cited publications were released before 2015. However, 12 of the top 20 publications regarding the DACY were published after 2015. This finding aligns with previous assertions regarding the inability of GCS to capture the scientific value or the publication’s influence solely.

3.2. Co-Authorship Analysis

Reconnoitering the scientific cooperation patterns facilitates access to expertise and allows widening the knowledge horizon [15,24]. The analysis and visualization of co-authorship networks provide an effective means to map these patterns. Consequently, the subsequent Section 3.2.1, Section 3.2.2, Section 3.2.3 present the co-authorship networks for countries, organizations, and authors, respectively.

3.2.1. Countries

Analyzing the countries’ scientific cooperation patterns can help in exploring the spatial distribution of AICJ publications and identifying the leading ones [16,20]. The co-authorship network for countries has been constructed utilizing VOSviewer, as illustrated in Figure 3. The thresholds for network processing are set at five for the number of documents per entity (NP), fifty for the number of citations per entity (CS), and a total link strength (TLS) threshold of 1. Consequently, 56 of the 80 countries that were analyzed meet the thresholds. The full details regarding the 56 countries are provided in

Table A1. AICJ countries’ cooperation.

Country	NP	CS	APY	ACD
China	1179	40,556	2018.2	34
USA	923	43,240	2015.6	47
South Korea	412	16,104	2016.6	39
Australia	287	14,928	2016.3	52
Canada	285	12,124	2017.2	43
England	282	13,369	2015.1	47
Taiwan	257	8948	2011.9	35
Singapore	158	6028	2018.7	38
Spain	153	4696	2017.5	31
Germany	128	4749	2018.3	37
Netherlands	106	2541	2016.4	24
Italy	100	3156	2018.4	32
Iran	84	2050	2019.0	24
Japan	82	2986	2014.2	36
Poland	77	1370	2015.2	18
Switzerland	63	2205	2018.1	35
France	58	2209	2015.9	38
Turkey	54	2214	2016.3	41
Portugal	47	2058	2014.9	44
Israel	46	2294	2012.5	50
Scotland	46	2089	2013.1	45
Finland	42	1501	2016.2	36
Belgium	36	1593	2019.0	44
Sweden	36	1424	2016.8	40
Egypt	36	1163	2015.6	32
New Zealand	28	825	2017.8	29
India	27	901	2017.6	33
Vietnam	26	1817	2019.8	70
U Arab Emirates	26	870	2019.0	33
Brazil	24	792	2018.4	33
Malaysia	23	634	2014.2	28
Wales	23	573	2016.9	25
Denmark	22	329	2021.3	15
Slovenia	21	711	2012.8	34
Ireland	20	684	2017.8	34
Chile	20	403	2017.3	20
Austria	20	197	2014.9	10
Saudi Arabia	18	471	2017.6	26
Greece	16	375	2014.0	23
Pakistan	11	420	2017.8	38
Lebanon	10	304	2017.6	30
Thailand	10	290	2011.4	29
Norway	10	168	2018.1	17
Serbia	9	200	2016.6	22
North Ireland	8	510	2018.4	64
Iraq	8	304	2020.6	38
Lithuania	8	287	2012.6	36
Indonesia	8	266	2017.3	33
Hungary	8	193	2017.5	24
Qatar	7	292	2018.7	42
Colombia	6	381	2015.2	64
Cyprus	6	134	2014.2	22
Turkey	6	70	2023.0	12
Luxembourg	5	287	2018.4	57
South Africa	5	233	2021.4	47
Jordan	5	227	2019.4	45

in Appendix A. In Figure 3, each country is represented by a distinct vertex, while the edges indicate the cooperation relations between countries, and their thickness reflects the degree of cooperation between countries with respect to the common documents. In Figure 3a, the size of the vertices represents the number of documents associated with each country, while the color variation denotes the authors’ average publication year (APY). The larger the vertex, the greater the research output, and the denser the red hue, the more recent publications. Conversely, in Figure 3b, the vertices’ size corresponds to the citations score of each country, while the color variation indicates the average citations per document (ACD) for those countries. Similarly, a larger vertex reflects higher citation score, and a dense yellow color represents countries with higher scientific value.

Table 4. AIC Leading Countries in Publication.

Country	Productivity	Influence	Activeness	Scientific Value
	NP	CS	APY	ACD
China	1179	40,556	2018	34
USA	923	43,240	2016	47
South Korea	412	16,104	2017	39
Australia	287	14,928	2016	52
Canada	285	12,124	2017	43
England	282	13,369	2015	47
Vietnam	26	1817	2020	70
Denmark	22	329	2021	15
North Ireland	8	510	2018	64
Iraq	8	304	2021	38
Colombia	6	381	2015	64
Türkiye	6	70	2023	12
Luxembourg	5	287	2018	57
South Africa	5	233	2021	47

Note: NP is number of documents per entity, CS is number of citations per entity, TLS total link strength (TLS), and ACD is average citations per document.

presents the leading countries in terms of NP, APY, CS, and ACD. Through Figure 3 and Table 4, the following findings are deduced:

• Concerning the publication productivity and impact, China (1179 publications, 40,556 citations) and the USA (923 publications, 43,240 citations) are the most productive and influential countries.
• Concerning activity, Turkey, South Africa, Iraq, Denmark, and Vietnam are the top active contributors with an APY of ≥2020.
• Regarding the document’s scientific value, Vietnam, Northern Ireland, Colombia, Luxembourg, and Australia have a robust impact, as evidenced by their ACD scores of ≥50 citations per document.

Regarding the cooperation intensity, the strongest cooperations are China–USA (131 mutual documents), China–Australia (111 mutual documents), USA–Republic of Korea (81 mutual documents), China–Singapore (70 mutual documents), China–England (61 mutual documents), USA–Canada (41 mutual documents), China–Republic of Korea (36 mutual documents), Republic of Korea–Australia (31 mutual documents), China–Canada (27 mutual documents), USA–Australia (26 mutual documents), and England–Australia (23 mutual documents). The occurrence of such intense collaborations, characterized by ≥20 mutual documents, is notably limited and predominantly observed among developed countries. This subset represents less than 3% of the overall collaborations, totaling 374 relations, as illustrated in Figure 3.

3.2.2. Organizations

Investigating the scientific cooperation of organizations can help support future academic partnerships and inform research policymaking among leading AICJ publishing organizations [15,25]. The co-authorship network for organizations is developed using VOSviewer, as depicted in Figure 4. The thresholds for NP, CS, and TLS in network processing are defined at 10, 100, and 1, respectively. Accordingly, 149 out of 2118 organizations meet the thresholds and are included in the analysis. The full details regarding the 149 organizations are provided in

Table A2. AICJ organizations’ cooperation.

Organization	NP	CS	APY	ACD
Hong Kong Polytech Univ	215	9436	2014	44
Tsinghua Univ	114	4536	2017	40
Huazhong Univ Sci and Technol	110	5610	2019	51
Georgia Inst Technol	94	7647	2014	81
Univ Alberta	87	2630	2018	30
Tongji Univ	87	2328	2020	27
Natl Taiwan Univ Sci and Technol	83	3135	2012	38
Natl Univ Singapore	74	2839	2018	38
City Univ Hong Kong	74	2778	2015	38
Yonsei Univ	73	2486	2017	34
Univ Hong Kong	72	3001	2017	42
Nanyang Technol Univ	63	2253	2019	36
Southeast Univ	61	1942	2021	32
Hong Kong Univ Sci and Technol	60	2668	2020	44
Natl Taiwan Univ	59	2005	2014	34
Curtin Univ	58	4212	2017	73
Univ Michigan	58	3053	2017	53
Concordia Univ	58	2793	2017	48
Univ Illinois	55	2812	2016	51
Hanyang Univ	55	2163	2017	39
Zhejiang Univ	54	1983	2018	37
Purdue Univ	48	1846	2014	38
Loughborough Univ	43	2907	2012	68
Univ Waterloo	43	2371	2016	55
Seoul Natl Univ	43	1190	2018	28
Kyung Hee Univ	42	2698	2014	64
Delft Univ Technol	42	711	2017	17
Penn State Univ	41	1787	2017	44
Carnegie Mellon Univ	39	2264	2012	58
Swiss Fed Inst Technol	39	1687	2018	43
Chongqing Univ	38	1054	2020	28
Chung Ang Univ	36	2451	2016	68
Univ Vigo	36	1232	2017	34
Texas A&M Univ	34	1560	2017	46
Univ Florida	34	1153	2019	34
Univ Cambridge	33	1606	2019	49
Harbin Inst Technol	32	1273	2018	40
Univ Southern Calif	31	3152	2016	102
Tianjin Univ	31	842	2019	27
RMIT Univ	30	2166	2018	72
Shenzhen Univ	30	1150	2020	38
Deakin Univ	29	2008	2016	69
Technion Israel Inst Technol	28	1603	2013	57
Dalian Univ Technol	28	1215	2020	43
Stanford Univ	27	901	2016	33
Univ Salford	26	1548	2007	60
Shanghai Jiao Tong Univ	26	1146	2019	44
Natl Cent Univ	26	896	2013	34
Wuhan Univ	26	788	2021	30
Korea Univ	26	627	2015	24
Eindhoven Univ Technol	26	571	2014	22
Univ Texas Austin	25	1049	2016	42
Natl Chiao Tung Univ	25	942	2009	38
Univ Twente	25	801	2017	32
Univ Tehran	25	512	2020	20
Queensland Univ Technol	24	1393	2015	58
UCL	24	718	2020	30
Univ British Columbia	24	628	2016	26
Tech Univ Munich	23	802	2019	35
Cairo Univ	22	838	2015	38
Sungkyunkwan Univ	22	641	2017	29
Univ Technol Sydney	21	609	2018	29
Univ Sydney	19	1006	2011	53
MIT	19	914	2012	48
Univ Toronto	19	681	2017	36
Changan Univ	19	594	2020	31
Monash Univ	19	515	2021	27
Virginia Tech	18	918	2017	51
Korea Inst Construct Technol	18	777	2010	43
Aalto Univ	18	746	2018	41
Natl Cheng Kung Univ	18	608	2008	34
Hohai Univ	18	537	2021	30
Osaka Univ	18	282	2010	16
Univ Washington	17	577	2016	34
Inha Univ	17	531	2012	31
Cardiff Univ	17	460	2018	27
Kyungpook Natl Univ	17	244	2020	14
Univ Ghent	16	1046	2018	65
Cent South Univ	16	505	2021	32
Amirkabir Univ Technol	16	441	2018	28
Univ Ljubljana	16	425	2012	27
Heriot Watt Univ	15	1388	2016	93
Hubei Engn Res Ctr Virtual Safe and Automated Const	15	1153	2019	77
Univ Auckland	15	459	2020	31
Northumbria Univ	15	415	2020	28
Politecn Torino	15	366	2015	24
Univ Seoul	15	330	2015	22
Univ Tokyo	15	301	2019	20
Univ Calif Berkeley	15	287	2011	19
Arizona State Univ	14	918	2014	66
Iran Univ Sci and Technol	14	506	2018	36
Univ Politecn Valencia	14	484	2018	35
Univ Salamanca	14	457	2019	33
Louisiana State Univ	14	427	2019	31
Sejong Univ	14	423	2019	30
Univ Strathclyde	14	423	2011	30
Griffith Univ	14	419	2011	30
Univ New South Wales	14	378	2020	27
Politecn Milan	14	334	2019	24
Southwest Jiaotong Univ	14	315	2021	23
Rhein Westfal Th Aachen	14	285	2019	20
Rutgers State Univ	13	869	2019	67
Columbia Univ	13	832	2016	64
Korea Adv Inst Sci and Technol	13	694	2019	53
Ruhr Univ Bochum	13	463	2019	36
Clemson Univ	13	453	2019	35
Singapore Univ Technol and Design	13	390	2019	30
Univ Calgary	13	332	2016	26
Chung Hua Univ	13	296	2009	23
Chinese Acad Sci	13	237	2022	18
Florida Int Univ	12	741	2011	62
NYU	12	693	2019	58
Univ Nebraska Lincoln	12	575	2017	48
Tech Univ Dresden	12	444	2021	37
Beijing Jiaotong Univ	12	408	2021	34
Shandong Univ	12	295	2020	25
South China Univ Technol	12	293	2021	24
Univ Granada	12	212	2020	18
Michigan State Univ	11	650	2016	59
Istanbul Tech Univ	11	587	2018	53
Northeastern Univ	11	523	2018	48
Univ Colorado	11	486	2016	44
Univ Nebraska	11	463	2017	42
Lulea Univ Technol	11	446	2015	41
Teesside Univ	11	438	2021	40
Univ Porto	11	435	2017	40
Imperial Coll London	11	428	2020	39
Natl Taipei Univ Technol	11	363	2014	33
Stevens Inst Technol	11	332	2019	30
Hunan Univ	11	318	2021	29
Hebei Univ Technol	11	306	2022	28
PCL Ind Management Inc	11	303	2019	28
Univ Wisconsin	11	259	2012	24
Seoul Natl Univ Sci and Technol	11	220	2018	20
Nanjing Tech Univ	11	189	2022	17
Jilin Univ	11	121	2021	11
Swinburne Univ Technol	10	597	2018	60
Univ Tecn Lisboa	10	480	2011	48
Univ Cent Florida	10	475	2012	48
Univ Alabama	10	459	2017	46
Univ Maryland	10	452	2013	45
Birmingham City Univ	10	447	2020	45
Univ Leeds	10	403	2018	40
Univ Lisbon	10	361	2017	36
Middle East Tech Univ	10	317	2021	32
IIT	10	300	2011	30
Tamkang Univ	10	243	2013	24
Univ Politecn Cataluna	10	127	2019	13
Univ Seville	10	122	2021	12

in Appendix A. Each organization is represented by a distinct vertex in Figure 4, and the edges indicate the collaborative relationships between organizations. The thickness of the edges indicates the level of cooperation between organizations, as measured by the number of mutual documents. As shown in Figure 4a, the size of the vertices represents the number of documents, whereas the color variation reflects the organization’s APY. The larger the vertex, the greater the research output, and the denser the red hue, the more recent publications. On the other hand, in Figure 4b, the size of the vertices represents the citation score of each organization, while the color variation represents the organizations’ ACD. Similarly, a larger vertex reflects a higher citation score, and a dense yellow color represents organizations with higher scientific value. Based on NP, APY, CS, and ACD, the pioneer organizations are listed in

Table 5. AICJ leading organizations.

Organization	Country	Productivity	Influence	Activeness	Scientific Value
		NP	CS	APY	ACD
Hong Kong Polytechnic University	China	215	9436	2014	44
Tsinghua University	China	114	4536	2017	40
Huazhong University of Science and Technology	China	110	5610	2019	51
Georgia Institute of Technology	USA	94	7647	2014	81
University of Alberta	Canada	87	2630	2018	30
Tongji University	China	87	2328	2020	27
Curtin University	Australia	58	4212	2017	73
University of Southern California	USA	31	3152	2016	102
Hohai University	China	18	537	2021	30
Heriot-Watt University	Scotland	15	1388	2016	93
Hubei Engineering Research Center for Virtual, Safe and Automated Construction	China	15	1153	2019	77
Southwest Jiaotong University	China	14	315	2021	23
Chinese Academy of Sciences	China	13	237	2022	18
Teesside University	England	11	438	2021	40
Hebei University of Technology	China	11	306	2022	28
Nanjing Tech University	China	11	189	2022	17

. Using Figure 4 and Table 5, the following findings are concluded:

• Concerning the document productivity and influence, Hong Kong Polytechnic University (215 documents, 9436 citations) is the top contributor in terms of document productivity and influence.
• Concerning activity, the Chinese Academy of Sciences, Hebei University of Technology, Nanjing Tech University, Hohai University, Teesside University, and Southwest Jiaotong University are the top active contributors with an APY > 2020.
• Regarding the document’s scientific value, the University of Southern California, Heriot-Watt University, Georgia Institute of Technology, Hubei Engineering Research Center for Virtual, Safe and Automated Construction, and Curtin University have the highest ACD scores with >70 citations per document, demonstrating the strong influence of their publications.
• Regarding the cooperation intensity, the strongest cooperations are Hong Kong Polytechnic University–Tsinghua University (18 mutual documents), Curtin University–Kyung Hee University (17 mutual documents), Hong Kong University of Science and Technology–National University of Singapore (17 mutual documents), Curtin University–Huazhong University of Science and Technology (15 mutual documents), Hong Kong Polytechnic University–Huazhong University of Science and Technology (15 mutual documents), Hong Kong Polytechnic University–Queensland University of Technology (15 mutual documents), Huazhong University of Science and Technology–Hubei Engineering Research Center for Virtual, Safe and Automated Construction (15 mutual documents), Huazhong University of Science and Technology–Nanyang Technological University (13 mutual documents), Hong Kong Polytechnic University–City University of Hong Kong (11 mutual documents), and Hong Kong Polytechnic University–University of Hong Kong (11 mutual documents). Figure 4 illustrates that there are 726 relations, and this type of significant organizational cooperation (≥10 mutual documents) accounts for less than 2% of the overall cooperation.
• It should be mentioned that twelve organizations in the United States and China are among the top organizations in Table 5. This aligns with the previous findings presented in Section 3.2.1, which indicate that China and the USA are the leading contributors in terms of total publication numbers and citation scores.

3.2.3. Authors

Authors are referenced as research producers. Hence, a detailed analysis of the authors’ co-authorship can characterize the leading ones while exploring their academic collaboration patterns. The co-authorship network for authors is established using VOSviewer, as illustrated in Figure 5. The thresholds established for network processing are as follows: NP is set at 10, CS at 100, and TLS at 1. Accordingly, 104 out of 9116 authors meet the thresholds and are included for analysis. The full details regarding the 104 authors are provided in

Table A3. AICJ authors’ cooperation.

Author	NP	CS	APY	ACD
Li, Heng	72	4187	2012	58
Cheng, Jack C. P.	48	2567	2020	53
Zhang, Limao	38	1209	2021	32
Ding, Lieyun	37	2942	2017	80
Haas, Carl T.	37	2134	2015	58
Cheng, Min-Yuan	34	1338	2011	39
Wang, Xiangyu	32	2578	2016	81
Luo, Han-Bin	31	2536	2018	82
Love, Peter E. D.	31	2462	2014	79
Al-Hussein, Mohamed	27	1078	2018	40
Teizer, Jochen	26	3528	2015	136
Kamat, Vineet R.	26	1317	2016	51
Eastman, Charles M.	21	1930	2015	92
Brilakis, Ioannis K.	21	1277	2016	61
Wang, Qian	21	1085	2020	52
Skitmore, Martin	21	1047	2016	50
Hammad, Amin	20	989	2017	49
Akinci, Burcu	19	1795	2012	94
Kim, Changwan	19	1589	2013	84
Chou, Jui-Sheng	19	928	2016	49
Lee, Ghang	19	755	2015	40
Seo, Jongwon	19	672	2016	35
Lee, Sanghyun	18	1527	2017	85
Sacks, Rafael	18	1475	2012	82
Lu, Weisheng	18	908	2018	50
Zhou, Cheng	18	749	2019	42
Taghaddos, Hossein	18	374	2021	21
Golparvar-Fard, Mani	17	1328	2018	78
Kim, Hyoungkwan	17	933	2017	55
Zhu, Zhenhua	17	833	2017	49
Abourizk, Simaan M.	17	520	2017	31
Wang, Shengwei	17	515	2012	30
Gan, Vincent J. L.	17	438	2022	26
Lee, Dong-Eun	17	256	2019	15
Zayed, Tarek	16	629	2017	39
Kang, Shih-Chung	16	597	2015	37
Cho, Yong K.	15	1036	2016	69
Moselhi, Osama	15	947	2015	63
Anumba, Chimay	15	864	2010	58
Ng, Shiu-Tong Thomas	15	824	2010	55
Edwards, David J.	15	621	2014	41
Arias, Pedro	15	609	2016	41
Tserng, H. Ping	15	595	2007	40
Liu, Jiepeng	15	154	2022	10
Son, Hyojoo	14	1151	2014	82
Zhou, Ying	14	1069	2017	76
Zhang, Jian-Ping	14	789	2010	56
Wang, Jun	14	678	2020	48
Luo, Xiao-Wei	14	567	2021	41
Bouferguene, Ahmed	14	472	2018	34
Chi, Seokho	14	434	2021	31
Bosche, Frederic	13	1288	2015	99
Luo, Xiaochun	13	918	2018	71
Pauwels, Pieter	13	734	2019	56
Wang, Mingzhu	13	625	2021	48
Zhang, Hong	13	623	2017	48
Cho, Hunhee	13	320	2014	25
Hong, Taehoon	13	164	2017	13
Park, Chan-Sik	12	898	2018	75
Xue, Fan	12	833	2020	69
Leite, Fernanda	12	635	2018	53
Wu, Yu-Wei	12	611	2014	51
Soibelman, Lucio	12	610	2013	51
Chi, Hung-Lin	12	549	2018	46
Shen, Qiping	12	529	2005	44
Dawood, Nashwan	12	527	2013	44
Rahimian, Farzad Pour	12	507	2020	42
Du, Jing	12	481	2021	40
Park, Moonseo	12	366	2015	31
Hermann, Ulrich	12	314	2019	26
Becerik-Gerber, Burcin	11	1374	2015	125
Matthews, Jane	11	670	2016	61
Chen, Jiayu	11	606	2019	55
Castro-Lacouture, Daniel	11	567	2012	52
De Soto, Borja Garcia	11	564	2020	51
Guo, Hongling	11	535	2016	49
Pan, Yue	11	517	2021	47
Arashpour, Mehrdad	11	495	2019	45
Hosseini, M. Reza	11	486	2020	44
Lu, Ming	11	451	2015	41
Marzouk, Mohamed	11	447	2018	41
Kim, Hongjo	11	427	2019	39
Han, Sanguk	11	371	2020	34
Li, Shuai	11	361	2020	33
Menassa, Carol C.	11	353	2019	32
Yu, Wen-Der	11	345	2010	31
El-Rayes, Khaled	11	342	2014	31
Mesnil, Romain	11	175	2020	16
Baverel, Olivier	11	153	2020	14
Zhong, Botao	10	811	2019	81
Li, Nan	10	714	2018	71
Wang, Wei-Chih	10	514	2009	51
Lam, Ka-Chi	10	443	2011	44
Cai, Hubo	10	433	2019	43
Koenig, Markus	10	394	2019	39
Lee, Yong-Cheol	10	378	2017	38
Ham, Youngjib	10	356	2020	36
Lee, Hyun-Soo	10	300	2016	30
Dzeng, Ren-Jye	10	290	2009	29
Vahdatikhaki, Faridaddin	10	274	2020	27
Ji, Ankang	10	270	2022	27
Chen, Ke	10	245	2021	25
Kang, Kyung-In	10	242	2013	24
Doree, Andre G.	10	205	2020	21

in Appendix A. In Figure 5, each vertex denotes a distinct author, whereas the edges linking the vertices show the cooperation relations between authors. The thickness of the edges indicates the intensity of the authors’ cooperation concerning the mutual documents. In contrast, the variation in vertex size and the coloring scheme depicted in Figure 5a,b are defined in accordance with Figure 3a,b, respectively.

Table 6. AICJ leading authors.

Author	Productivity	Influence	Activeness	Scientific Value
	NP	CS	APY	ACD
Heng Li	72	4187	2012	58
Jack C. P. Cheng	48	2567	2020	53
Limao Zhang	38	1209	2021	32
Lieyun Ding	37	2942	2017	80
Carl T. Haas	37	2134	2015	58
Xiangyu Wang	32	2578	2016	81
Jochen Teizer	26	3528	2015	136
Charles M. Eastman	21	1930	2015	92
Burcu Akinci	19	1795	2012	94
Vincent J. L. Gan	17	438	2022	26
Jiepeng Liu	15	154	2022	10
Frederic Bosche	13	1288	2015	99
Burcin Becerik-Gerber	11	1374	2015	125
Yue Pan	11	517	2021	47
Ankang Ji	10	270	2022	27

comprises the pioneer authors in terms of NP, APY, CS, and ACD. Through Figure 5 and Table 6, the following findings are deduced:

• Concerning the productivity and impact of the documents, Heng Li is observed as the most productive cited author with a total of 72 publications, and the number of citations is 4187. Li Heng’s research interests involve construction workers’ safety, ergonomics, deep learning, computer vision, robotics, building information modeling, and wearables [26,27,28,29,30,31,32].
• As for the significance of the document, the ACD score of Jochen Teizer is the highest, with 136 citations per document. His areas of interest in research involve photogrammetry, Robotic Total Station (RTS), building information modeling, safety, laser scanning, and construction equipment operation [33,34,35,36,37,38,39,40].
• Concerning the activeness, Ankang Ji, Yue Pan, Limao Zhang, Vincent J. L. Gan, and Jiepeng Liu are the top active authors with APY ≥ 2021. Their research interests are related to defect detection, crack detection, image segmentation, rebar layout, building information modeling (BIM), digital twin, modular construction, terrestrial laser scanning (TLS), 3D point cloud, social network analysis (SNA), evacuation modeling, and structural safety [41,42,43,44,45,46,47].

3.3. Science Mapping

Author-keywords (AKs) are the phrases that represent the central topic or core content of publications. Accordingly, exploring AKs allow us to characterize the focal interests and hotspots for a given research entity [25,48]. In this light, AKs are analyzed and explored based on their co-occurrence frequency and their temporal evolution for AICJ publications as per Section 3.3.1 and Section 3.3.2, respectively.

3.3.1. Co-Occurrence Analysis

Using VOSviewer, the co-occurrence network is generated as shown in Figure 6. For network processing, the threshold for AKs’ frequency is set at 10 while a thesaurus file is developed and employed for merging the identical AKs (e.g., ‘ANN’ to ‘Artificial Neural Networks’, ‘BIM’ to ‘Building Information Modeling’, etc.) to avoid potential redundancy or findings disruption. Accordingly, 219 out of 9223 keywords meet the threshold and included for analysis. The full details regarding the 219 AKs are provided in

Table A4. AICJ author-keyword co-occurrence.

Author-Keyword	Cluster	Frequency	APY
building information modeling (BIM)	3	598	2018
deep learning (DL)	2	232	2022
point cloud	2	117	2020
machine learning (ML)	5	114	2020
computer vision	2	114	2019
industry foundation classes (IFCs)	3	112	2017
construction	4	110	2012
optimization	1	107	2016
genetic algorithm (GA)	1	104	2014
convolutional neural network (CNN)	2	99	2021
automation	4	95	2015
simulation	1	91	2013
artificial neural network (ANN)	5	81	2015
virtual reality (VR)	5	80	2016
scheduling	1	76	2014
construction safety	5	75	2019
robotics	4	71	2017
construction engineering and management	1	71	2013
unmanned aerial vehicle (UAV)	2	69	2021
ontology	3	66	2017
structural health monitoring (SHM)	2	58	2021
3D printing	4	56	2020
laser scanning	2	53	2016
image processing	2	52	2016
augmented reality (AR)	3	52	2016
artificial intelligence (AI)	5	51	2018
visualization	3	51	2013
digital twin	3	48	2022
computer-aided design	1	48	2009
internet of things (IoT)	3	46	2020
architecture-engineering-construction	3	46	2015
interoperability	3	45	2016
semantic segmentation	2	44	2022
multi-objective optimization	1	44	2018
information and communication technology (ICT)	4	44	2009
progress monitoring/tracking	2	43	2018
fuzzy logic	1	43	2012
planning	1	42	2014
radio frequency identification (RFID)	4	42	2012
facilities management (FM)	3	41	2017
safety	4	41	2015
project management	4	40	2014
3D modeling	1	40	2013
object detection	2	38	2021
productivity	4	38	2015
ground penetrating radar (GPR)	2	37	2019
construction machines	5	37	2017
terrestrial laser scanning (TLS)	2	36	2020
support vector machine (SVM)	5	36	2016
decision support system (DSS)	1	36	2013
geographic information system (GIS)	3	35	2015
construction automation	4	35	2014
modular construction	3	34	2021
semantic web	3	34	2019
finite element modeling	2	34	2016
blockchain technology	3	33	2022
3D reconstruction	2	33	2020
light detection and ranging (lidar)	2	33	2020
non-destructive testing (NDT)	2	32	2020
site layout planning	1	32	2014
path planning	4	30	2018
earthmoving operations/projects	4	30	2014
natural language processing (NLP)	5	29	2021
crack detection	2	29	2020
agent based modeling	1	29	2018
decision-making	1	29	2015
construction site	2	28	2017
parametric design (PD)	1	28	2017
quality control	4	28	2016
smart contracts	3	27	2021
generative design	1	27	2019
excavator	5	27	2017
particle swarm optimization (PSO)	1	27	2015
construction projects	3	26	2014
safety management	5	25	2018
photogrammetry	2	25	2016
knowledge management (KM)	3	25	2011
additive manufacturing	4	24	2020
concrete	4	24	2016
data mining	5	24	2016
case-based reasoning (CBR)	5	24	2013
linked data	3	23	2019
digital fabrication	4	23	2018
inspection	2	23	2016
collaboration	3	23	2011
construction worker	5	22	2018
real-time monitoring	4	22	2017
tower cranes	1	22	2017
bridge inspection	2	22	2017
segmentation	2	21	2017
integration	3	21	2012
long short-term memory (LSTM)	2	20	2021
prefabrication	3	20	2016
multi-criteria decision analysis (MCDA)	1	20	2013
reinforcement learning	1	19	2022
tunneling	5	19	2018
maintenance	1	19	2017
uncertainty	1	19	2016
multi-agent simulation	3	19	2013
building	4	19	2012
information management	4	19	2010
global positioning system (GPS)	4	19	2010
collaborative design	3	19	2006
pose estimation	5	18	2020
data fusion	5	18	2018
project control	5	18	2014
4D CAD	1	18	2010
random forest	5	17	2020
masonry	2	17	2020
energy efficiency	3	17	2019
infrared thermography (IRT)	2	17	2019
reinforcement steel bars	4	17	2018
lean construction	3	17	2017
clustering	2	17	2017
risk management	3	17	2015
resource-constrained project scheduling problem (RCPSP)	1	17	2014
performance	3	17	2012
offsite construction	3	16	2021
defect detection	2	16	2020
wearables	5	16	2019
bridges	1	16	2017
classification	2	16	2017
sustainability	1	16	2017
heritage buildings	2	16	2017
discrete event simulation (DES)	1	16	2015
cranes	4	16	2014
decision support	3	16	2014
design process	1	16	2012
architectural design	1	16	2011
architecture	1	16	2009
tunnel boring machine (TBM)	5	15	2021
transfer learning	2	15	2021
computational design	3	15	2019
structure-from-motion	2	15	2019
sewer	2	15	2018
metaheuristics	1	15	2018
social network analysis (SNA)	3	15	2017
text mining	5	15	2016
energy saving	5	15	2016
conceptual design	1	15	2014
sensors	4	15	2013
virtual prototyping	1	15	2012
process modeling	1	15	2010
generative adversarial network	2	14	2022
activity recognition	5	14	2021
electroencephalography (EEG)	5	14	2020
scan-to-BIM	2	14	2020
localization	4	14	2020
parametric modeling	1	14	2019
indoor localization	2	14	2018
algorithms	1	14	2018
infrastructure	1	14	2018
mass customization (mc)	1	14	2017
real time	4	14	2016
supply chain management (SCM)	3	14	2016
Monte Carlo simulation	1	14	2015
3D visualization	1	14	2015
linear scheduling	1	14	2014
building management system (BMS)	3	14	2013
modeling	1	14	2011
resource assignment	4	14	2010
information systems	4	14	2008
instance segmentation	2	13	2022
shield tunneling	2	13	2020
hydraulic excavator	1	13	2020
mobile laser scanning (MLS)	2	13	2019
sensitivity analysis	1	13	2018
structural design	1	13	2016
heuristics	1	13	2015
3D laser scanning	2	13	2015
wireless sensor network (WSN)	4	13	2015
building automation systems	3	13	2014
cost estimating	3	13	2014
case study	5	13	2014
monitoring	4	13	2012
evaluation	4	13	2008
internet	3	13	2004
data integration	3	12	2021
robotic fabrication	4	12	2021
asset management	3	12	2020
evolutionary algorithms	1	12	2019
damage detection	2	12	2019
condition assessment	2	12	2018
occupational health and safety (OHS)	5	12	2018
repetitive projects	1	12	2015
benchmarking	3	12	2014
tracking	4	12	2013
costs	1	12	2013
shape grammars	1	12	2012
analytic hierarchy process (AHP)	1	12	2011
design	1	12	2011
databases	3	12	2009
product modeling	1	12	2008
attention mechanism	2	11	2023
digitalization	3	11	2022
industry 4.0	3	11	2022
asphalt pavement	2	11	2021
image segmentation	2	11	2021
prefabricated construction	3	11	2020
cyber–physical system (CPS)	5	11	2020
stereovision	2	11	2020
human–robot collaboration (HRC)	4	11	2019
as-built modeling	2	11	2018
ergonomics	5	11	2018
intelligent compaction	4	11	2018
design automation	1	11	2017
cloud computing	3	11	2016
remote sensing	2	11	2016
HVAC systems	3	11	2016
tunnels	2	11	2016
health and safety (H&S)	4	11	2015
prototype	1	11	2014
machine vision	4	11	2014
life cycle costing (LCC)	1	11	2014
ant colony optimization	1	11	2013
precast concrete	1	11	2013
ready mixed concrete (RMC)	4	11	2012
rapid prototyping	1	11	2006
data augmentation	2	10	2022
synthetic dataset	2	10	2021
vision-based	2	10	2020
point cloud processing	2	10	2020
life cycle assessment (LCA)	1	10	2019
inertial measurement unit (IMU)	5	10	2019
high-rise buildings	1	10	2019
big data	3	10	2019
geometry	3	10	2019
risk assessment	5	10	2018
model view definition (MVD)	3	10	2018
quantity take-off	3	10	2017
rule checking	3	10	2017
teleoperation	5	10	2017
BIM adoption	3	10	2016
delays	5	10	2016
fault detection and diagnosis (FDD)	3	10	2016
underground construction	2	10	2016
building simulation	1	10	2016
safety risks	5	10	2015
ultra-wide band (UWB) technology	4	10	2015
principal component analysis (PCA)	3	10	2015
constraint programming	1	10	2013
object recognition	2	10	2013
construction simulation	5	10	2012
expert systems	1	10	2011
construction materials	4	10	2011

in Appendix A. In Figure 6, the AKs are represented by the vertices, and the co-occurrence relations between them are represented by the edges. The thickness of these edges indicates the intensity of the co-occurrence between the AKs in terms of the mutual documents. On the other hand, the vertices’ size refers to the AKs’ frequency, while the color variation refers to the AKs’ clusters. In Figure 6a, the vertices’ size denotes the AKs’ frequency while the variation of color refers to the AKs’ average publication year (APY). On the other hand, in Figure 6b, the vertices’ size denotes the AKs’ frequency while the color variation refers to the AKs’ clusters.

Table 7. Top AKs of AICJ publications.

AK	Cluster	Frequency	APY
Building information modeling (BIM)		598	2018.30
Deep learning (DL)		232	2021.61
Point cloud		117	2019.98
Machine learning (ML)		114	2020.34
Computer vision		114	2019.26
Industry foundation classes (IFCs)		112	2017.25
Construction		110	2012.34
Optimization		107	2016.02
Genetic algorithm (GA)		104	2014.17
Convolutional neural network (CNN)		99	2020.74
Automation		95	2014.65
Simulation		91	2012.95
Artificial neural network (ANN		81	2015.00
Virtual reality (VR)		80	2016.05
Scheduling		76	2013.72
Construction safety		75	2018.68
Robotics		71	2016.61
Construction engineering and management		71	2013.41
Unmanned aerial vehicle (UAV)		69	2020.57
Ontology		66	2016.61
Digital twin		48	2022.13
Semantic segmentation		44	2022.41
Blockchain technology		33	2021.58
Natural language processing (NLP)		29	2020.97
Smart contracts		27	2021.48
Long short-term memory (LSTM)		20	2020.95
Reinforcement learning		19	2022.05
Offsite construction		16	2021.44
Tunnel boring machine (TBM)		15	2021.13
Transfer learning		15	2020.93
Generative adversarial network		14	2021.86
Activity recognition		14	2021.07
Instance segmentation		13	2022.15
Attention mechanism		11	2022.55
Digitalization		11	2021.91
Industry 4.0		11	2021.73
Asphalt pavement		11	2021.09
Data augmentation		10	2021.70

comprises the top AKs in terms of frequency and APY. According to Figure 6b, five major scientific clusters or communities are detected for the AICJ applications.

3.3.2. Thematic Evolution

Despite the fact that Figure 6 and Table 7 provide significant insights into the current state of AKs, they are unable to accurately depict or demonstrate their development and distribution over time. For this purpose, the thematic evolution function in Biblioshiny is utilized. As shown in Figure 7, this function is grounded by the principles of social network analysis (SNA) to analyze and visualize the evolution of author-keywords, which represent the focal topics of AICJ publications, across multiple subsequent time-slice maps while providing insights into their structural changes and trends. In each time-slice map, the associated AKs are categorized into distinct thematic areas based on their co-occurrence relationships. Then, a 2D visualization of each thematic area is provided with a definite label and size and located according to its Callon centrality (X-axis) and Callon density (Y-axis) [49,50]. The size of the circle (i.e., diameter) is proportional to the overall frequencies of all AKs in the theme, whereas the label refers to the theme’s most commonly recurring AK. Callon density measures the thematic area’s development status, whereas Callon Centrality measures the theme or thematic area’s significance or relevance. As a result, the themes are divided into the following four typologies.

(a)

Motor themes in the upper right quadrant can be identified by their high density and high centrality, indicating that they are thoroughly established and significant for the field of research,

(b)

Basic themes in the lower-right quadrant’s can be identified by their low density and high centrality, indicating their significance for the field of study and their concern with general topics that are related to different research areas in the field,

(c)

Niche themes in the upper-left quadrant can be identified by their high density and low centrality, indicating their high level of development and relative lack of significance for the research field, and

(d)

Emerging or declining themes in the lower-left quadrant can be identified by their low centrality and low density, indicating that these themes are either emerging/arising or weakly developed/marginal.

To crystallize the main AICJ research themes over time, it is decided to divide the temporal interval of extracted data into four time-slices as per Section 3.1.1 while using Biblioshiny recommended settings: Clustering Algorithm “ Walktrap”, Maximum Number of keywords “250”, Minimum Cluster Frequency “0.005”, Weight Index “Inclusion Index weighted by Word-Occurrence”, and Min Weight Index “0.1”.

• 2000–2005 Time-Slice

In the 2000–2005 time-slice (Figure 7a), there are 322 published documents, distributed in the four typologies as follows:

(a)

Ten thematic areas belong to motor themes [global positioning system (GPS), integration, automation, construction processes, computer-integrated construction, architecture, industry foundation classes (IFCs), document management, e-commerce (EC), and decision support system (DSS)]. These themes represent the initial phase of digital transformation in the construction industry, focusing on productivity, interoperability, and data flow. GPS and automation enhanced site operations and equipment tracking, while frameworks such as IFC and computer-integrated construction (CIC) facilitated improved coordination across project stages. The presence of DSS and e-commerce highlights early efforts to use digital tools for decision-making and supply chain management.

Table A5. Details for 2000–2005 motor themes.

Cluster Label	Related Topics	Occurrences
architecture	analytic hierarchy process (AHP)	2
	architectural design	4
	architecture	8
	artificial intelligence (AI)	2
	assemblies	2
	collaboration	6
	CSCW	3
	design	5
	design education	3
	design studio	2
	environmental design	2
	heuristics	2
	optimization	6
	rapid prototyping	6
	virtual studio	2
automation	automation	10
	benchmarking	2
	computer vision	3
	construction automation	3
	earthmoving operations/projects	4
	intelligent systems	2
	key performance indicators (KPIs)	2
	laser	2
	performance	2
	resource assignment	2
	robotics	6
	sensors	2
	tile setting	2
computer-integrated construction	computer-integrated construction	2
	design automation	2
construction processes	computer simulation	2
	construction component	2
	construction processes	3
	verification and validation	2
decision support system (DSS)	artificial neural network (ANN)	6
	data mining	2
	data warehouse	3
	decision support system (DSS)	7
	enterprise resource planning (ERP)	2
	expert systems	2
	fuzzy logic	4
	geographic information system (GIS)	5
	online analytical processing (OLAP)	2
	project management	2
	safety monitoring	2
	site layout planning	6
document management	document management	3
	facilities management (FM)	2
	unified modeling language	2
e-commerce (EC)	construction materials	3
	e-commerce (EC)	4
	intelligent agent (IA)	4
	multi-agent simulation	3
	procurement	3
	supply chain management (SCM)	2
	xml	3
global positioning system (GPS)	automated data collection (ADC)	2
	control	3
	global positioning system (GPS)	5
	monitoring	3
	road construction	2
industry foundation classes (IFCs)	4D CAD	2
	implementation	2
	industry foundation classes (IFCs)	5
	international alliance for interoperability (IAI)	2
	knowledge-based systems	3
integration	integration	3
	interoperability	2
	local area network (LAN)	2
	web services	2

of Appendix B illustrates the inner related-topics in each motor theme in 2000-2005 time-slice.

(b)

Six thematic areas belong to basic themes [information and communication technology (ICT), virtual environments, construction, genetic algorithms, conceptual design, and data modeling].

Table A6. Details for 2000–2005 basic themes.

Cluster Label	Related Topics	Occurrences
conceptual design	conceptual design	3
	shape grammars	3
construction	agents	2
	building	5
	computer	2
	computer-aided design	19
	construction	23
	design process	4
	digital libraries	2
	knowledge management (KM)	5
	product modeling	3
	reuse	2
	space	2
data modeling	data modeling	3
genetic algorithm (GA)	4D site management	3
	case-based reasoning (CBR)	3
	computational fluid dynamics (CFD)	2
	decision support	3
	genetic algorithm (GA)	13
	infrastructure management	2
	object-oriented	2
	scheduling	10
	visualization	10
information and communication technology (ICT)	3D modeling	6
	collaborative design	13
	constructability	2
	construction engineering and management	5
	construction projects	5
	construction simulation	3
	costs	2
	discrete event simulation (DES)	2
	evaluation	4
	hypermedia	2
	information and communication technology (ICT)	15
	information exchange	2
	information management	4
	information systems	5
	management information systems	2
	modeling	2
	performance measurement	4
	planning	8
	process improvement strategies	2
	process modeling	5
	production rate	2
	simulation	9
	virtual reality (VR)	13
	VRML	5
virtual environments	design collaboration	2
	virtual environments	4
	virtual worlds	3

of Appendix B illustrates the inner related topics in each basic theme in the 2000–2005 time-slice.

(c)

Six thematic areas belong to niche themes [architectural modeling, geometric modeling, knowledge representation, operation, internet, and autonomous vehicles].

Table A7. Details for 2000–2005 niche themes.

Cluster Label	Related Topics	Occurrences
architectural modeling	architectural modeling	2
	computer-supported cooperative work (CSCW)	2
	concept modeling	2
	concurrent engineering	2
	product data modeling	2
	virtual prototyping	2
autonomous vehicles	autonomous vehicles	3
	barcodes	2
	image processing	3
	tunneling	2
geometric modeling	geometric modeling	2
	parametric design (PD)	2
internet	building management system (BMS)	4
	databases	4
	education	2
	energy	2
	extranet	2
	internet	10
	learning	2
	multimedia	3
	real estate	2
	World Wide Web	2
knowledge representation	design context	2
	knowledge representation	3
	sustainability	2
operation	feedforward	2
	interface design	2
	operation	3

of Appendix B illustrates the inner related topics in each niche theme in the 2000-2005 time-slice.

(d)

Two thematic areas belong to emerging themes [architecture–engineering–construction, and construction site].

Table A8. Details for 2000–2005 emerging themes.

Cluster Label	Related Topics	Occurrences
architecture–engineering–construction	architecture–engineering–construction	5
	standardization	2
construction site	construction site	3
	space layout planning	2

of Appendix B illustrates the inner related topics in each emerging theme in the 2000–2005 time-slice.

• 2006–2011 Time-Slice

In the 2006–2011 time-slice (Figure 7b), there are 536 published documents, distributed in the four typologies as follows:

(a)

Eight thematic areas belong to motor themes [multi-criteria decision analysis (MCDA), image processing, contractor, architectural design, business process reengineering (BPR), decision-making, case study, radio frequency identification (RFID)]. These themes reflect a focus on decision support, automated data capture, and process improvement. MCDA and decision-making highlight efforts to enhance planning through complex evaluations, while image processing and RFID signal progress in real-time data tracking. Contractor, architectural design, and BPR emphasize the drive to optimize project delivery through technology and organizational reform.

Table A9. Details for 2006–2011 motor themes.

Cluster Label	Related Topics	Occurrences
architectural design	architectural design	4
	heritage buildings	3
	rapid prototyping	3
	computer-aided architectural design (CAAD)	2
	computer modeling	2
business process reengineering (BPR)	business process reengineering (BPR)	3
	design build	3
	collaborative design	2
	intelligent agent (IA)	2
	internet	2
case study	case study	4
	discrete event simulation	4
	validation	4
	verification and validation	4
	animation	2
	business processes	2
	design review	2
	economic analysis	2
contractor	contractor	3
	structural equation modeling (SEM)	3
	contract negotiation	2
	enterprise resource planning (ERP)	2
	markup	2
decision-making	decision-making	7
	prefabrication	5
	design process	4
	concrete construction	3
	design evaluation	3
	generative design	3
	construction methods	2
	design knowledge	2
	multi-attribute utility	2
image processing	image processing	10
	automated inspection	3
	pipeline	3
	segmentation	3
	crack detection	2
	mathematical morphology	2
multi-criteria decision analysis (MCDA)	multi-criteria decision analysis (MCDA)	6
	underground construction	3
	analytic network process (ANP)	2
	environmental assessment	2
	environmental impact	2
	expert knowledge	2
radio frequency identification (RFID)	radio frequency identification (RFID)	19
	augmented reality (AR)	8
	construction automation	8
	laser scanning	8
	safety	7
	construction machines	3
	human factors	3
	path planning	3
	remote control	3
	stability	3
	ultra-wide band (UWB) technology	3
	3D	2
	blind spots	2
	building automation	2
	location tracking and monitoring	2
	mixed reality (MR)	2

of Appendix B illustrates the inner related topics in each motor theme in the 2006–2011 time-slice.

(b)

Nine thematic areas belong to basic themes [knowledge management (KM), multi-agent simulation, construction site, intelligent buildings, support vector machine (SVM), 3D modeling, artificial neural network (ANN), construction engineering and management, and building information modeling (BIM)].

Table A10. Details for 2006–2011 basic themes.

Cluster Label	Related Topics	Occurrences
3D modeling	3D modeling	10
	4D CAD	9
	ontology	8
	earthmoving operations/projects	5
	product modeling	5
	workflow	5
	linear scheduling	4
	process modeling	4
	3D CAD	3
	3D imaging	3
	clustering	3
	virtual prototyping	3
	automated object recognition	2
	constructability	2
	infrastructure projects	2
	intelligent excavation system	2
artificial neural network (ANN)	artificial neural network (ANN)	20
	fuzzy logic	17
	information and communication technology (ICT)	15
	architecture–engineering–construction	12
	artificial intelligence (AI)	8
	concrete	8
	case-based reasoning (CBR)	7
	modeling	6
	evaluation	5
	construction supply chain	3
	data mining	3
	k-means	3
	neuro-fuzzy	3
	PDA	3
	performance	3
	subcontractors	3
	XML	3
	briefing	2
	cash flow	2
	change orders	2
	design management	2
	disputes	2
	litigation	2
building information modeling (BIM)	building information modeling (BIM)	38
	construction	32
	automation	23
	project management	15
	computer-aided design	13
	visualization	12
	industry foundation classes (IFCs)	10
	information management	10
	robotics	10
	interoperability	9
	productivity	9
	resource assignment	8
	global positioning system (GPS)	7
	inspection	7
	collaboration	6
	integration	6
	photogrammetry	6
	progress monitoring/tracking	6
	project control	6
	quality control	6
	ready-mixed concrete (RMC)	6
	computer vision	5
	technology	5
	3D visualization	4
	bridge inspection	4
	cost estimating	4
	geographic information system (GIS)	4
	machine vision	4
	tracking	4
	3D laser scanning	3
	3D object modeling	3
	automated data collection (ADC)	3
	building	3
	communication	3
	facilities management (FM)	3
	location sensing	3
	object detection	3
	prototype	3
	sensors	3
	tunnels	3
	WLAN	3
	agents	2
	bill of quantity (BOQ)	2
	change management	2
	conceptual design	2
	construction project data	2
	data acquisition	2
	documentation	2
	emulation	2
	expert systems	2
	geometric modeling	2
	highway projects	2
	images	2
	labor	2
construction engineering and management	construction engineering and management	23
	simulation	23
	genetic algorithm (GA)	21
	scheduling	18
	optimization	14
	planning	7
	particle swarm optimization (PSO)	6
	performance evaluation	6
	resource-constrained project scheduling problem (RCPSP)	6
	ant colony optimization	5
	multi-objective optimization	5
	system dynamics	5
	analytic hierarchy process (AHP)	4
	constraint programming	4
	critical path method	4
	risk management	4
	bridge maintenance	3
	construction materials	3
	costs	3
	data fusion	3
	infrastructure	3
	life cycle costing (LCC)	3
	materials tracking	3
	Monte Carlo simulation	3
	repetitive projects	3
	stochastic time-cost	3
	uncertainty	3
	acceleration	2
	conflict analysis	2
	construction management process reengineering	2
	construction safety	2
	dynamic site layout planning	2
	e-commerce (EC)	2
	fiber-reinforced materials	2
	housing	2
	location estimation	2
	max–min ant system	2
	metaheuristics	2
	models	2
construction site	construction site	4
intelligent buildings	intelligent buildings	4
	building automation systems	3
	web services	3
	model	2
knowledge management (KM)	knowledge management (KM)	9
	information systems	5
	construction projects	4
	databases	3
	knowledge map	3
	value engineering	3
	web-based application	3
multi-agent simulation	multi-agent simulation	3
support vector machine (SVM)	support vector machine (SVM)	9
	decision support system (DSS)	7
	fast messy genetic algorithms	4
	contractor selection	3
	housing refurbishment	2
	multiple-criteria analysis	2

of Appendix B illustrates the inner related topics in each basic theme in the 2006–2011 time-slice.

(c)

Five thematic areas belong to niche themes [benchmarking, evolutionary strategies, finite element modeling, monitoring, and web-based systems].

Table A11. Details for 2006–2011 niche themes.

Cluster Label	Related Topics	Occurrences
benchmarking	benchmarking	2
	Hong Kong	2
	key performance indicators (KPIs)	2
evolutionary strategies	evolutionary strategies	2
	lean construction	2
	lean production	2
finite element modeling	finite element modeling	9
	adiabatic hydration curves	2
	computer simulation	2
	formwork	2
monitoring	monitoring	3
	control methods	2
	data collection	2
	feedback control	2
web-based system	web-based system	4
	decision support	3
	delays	3
	construction claims	2
	factor selection	2

of Appendix B illustrates the inner related topics in each niche theme in the 2006–2011 time-slice.

(d)

Three thematic areas belong to emerging themes [building simulation, excavator, and virtual reality (VR)].

Table A12. Details for 2006–2011 emerging themes.

Cluster Label	Related Topics	Occurrences
building simulation	building simulation	4
excavator	excavator	3
	hybrid systems	3
virtual reality (VR)	virtual reality (VR)	8
	cranes	6
	equipment and machinery	4
	health and safety (H&S)	3
	construction simulation	2

of Appendix B illustrates the inner related topics in each emerging theme in the 2006–2011 time-slice.

• 2012–2017 Time-Slice

In the 2012–2017 time-slice (Figure 7c), there are 1001 published documents, distributed in the four typologies as follows:

(a)

Two thematic areas belong to motor themes [finite element modeling and laser scanning]. The emergence of laser scanning as a motor theme during the 2012–2017 period can be attributed to the convergence of technological advancements and evolving industry needs. Notable improvements in data processing capabilities, along with the increased availability and affordability of terrestrial laser scanners, enabled construction professionals to efficiently capture high-precision as-built data. This technological advancement was closely linked to the increasing use of simulation and analytical tools in construction engineering, where finite element modeling (FEM) played a critical role in predicting structural behavior, optimizing design alternatives, and enhancing safety and performance assessments across complex construction scenarios.

Table A13. Details for 2012–2017 motor themes.

Cluster Label	Related Topics	Occurrences
finite element modeling	finite element modeling	6
	segmentation	6
	light detection and ranging (lidar)	5
	terrestrial laser scanning (TLS)	5
laser scanning	laser scanning	22
	point cloud	19
	automation	16
	computer vision	15
	construction machines	14
	robotics	14
	progress monitoring/tracking	11
	earthmoving operations/projects	10
	construction automation	9
	image processing	9
	path planning	9
	3D reconstruction	8
	construction worker	8
	photogrammetry	8
	bridge inspection	7
	cranes	7
	ground-penetrating radar (GPR)	7
	non-destructive testing (NDT)	7
	unmanned aerial vehicle (UAV)	7
	virtual prototyping	7
	3D laser scanning	6
	clustering	6
	global positioning system (GPS)	6
	infrared thermography (IRT)	6
	object recognition	6
	tracking	6
	3D data	5
	Hough transform	5
	project control	5
	remote sensing	5

of Appendix B illustrates the inner related topics in each motor theme in the 2012–2017 time-slice.

(b)

Six thematic areas belong to basic themes [3D printing, safety, building information modeling (BIM), optimization, radio frequency identification (RFID), and uncertainty].

Table A14. Details for 2012–2017 basic themes.

Cluster Label	Related Topics	Occurrences
building information modeling (BIM)	building information modeling (BIM)	173
	simulation	34
	industry foundation classes (IFCs)	31
	ontology	27
	construction engineering and management	25
	scheduling	24
	construction	22
	construction safety	22
	augmented reality (AR)	21
	interoperability	17
	visualization	15
	facilities management (FM)	12
	planning	11
	project management	11
	semantic web	10
	geographic information system (GIS)	9
	construction projects	8
	information and communication technology (ICT)	8
	linked data	8
	multi-agent simulation	8
	risk management	8
	social network analysis (SNA)	8
	4D CAD	7
	BIM adoption	6
	cloud computing	6
	cost estimating	6
	integration	6
	internet of things (IoT)	6
	lean construction	6
	prototype	6
	building	5
	discrete event simulation (DES)	5
	energy efficiency	5
	measurements	5
	mobile computing	5
	quantity take-off	5
	supply chain management (SCM)	5
	sustainability	5
optimization	optimization	35
	genetic algorithm (GA)	32
	artificial neural network (ANN)	20
	site layout planning	14
	support vector machine (SVM)	14
	fuzzy logic	13
	machine learning (ML)	12
	multi-objective optimization	12
	particle swarm optimization (PSO)	12
	decision support system (DSS)	11
	data mining	9
	computer-aided design	8
	tower cranes	8
	artificial intelligence (AI)	6
	bridges	6
	case-based reasoning (CBR)	6
	knowledge management (KM)	6
	expert systems	5
	maintenance	5
	regression analysis	5
radio frequency identification (RFID)	radio frequency identification (RFID)	17
	real-time monitoring	10
	wireless sensor network (WSN)	8
	quality control	7
	quality assessment	5
safety	safety	19
	productivity	16
	virtual reality (VR)	11
	3D modeling	10
	decision-making	10
	agent-based modeling	8
	architecture–engineering–construction	8
	performance	8
	safety management	7
	wearables	7
	occupational health and safety (OHS)	6
	safety risks	6
	sensors	6
	activity analysis	5
	indoor localization	5
	real-time location estimation systems (RTLSs)	5
uncertainty	uncertainty	9
	multi-criteria decision analysis (MCDA)	8
3D printing	3D printing	7
	digital fabrication	6

of Appendix B illustrate the inner related topics in each basic theme in the 2012–2017 time-slice.

(c)

Five thematic areas belong to niche themes [excavator, HVAC systems, linear scheduling, technology acceptance model, and structural optimization].

Table A15. Details for 2012–2017 niche themes.

Cluster Label	Related Topics	Occurrences
excavator	excavator	9
	energy recovery	5
HVAC systems	HVAC systems	7
	principal component analysis (PCA)	5
linear scheduling	linear scheduling	6
	repetitive projects	5
	resource leveling	5
structural optimization	structural optimization	7
	structural design	5
technology acceptance model (TAM)	technology acceptance model (TAM)	5

of Appendix B illustrates the inner related topics in each niche theme in the 2012–2017 time-slice.

(d)

Three thematic areas belong to emerging themes [energy saving, Monte Carlo simulation, and parametric design (PD)].

Table A16. Details for 2012-2017 emerging themes.

Cluster Label	Related Topics	Occurrences
energy saving	energy saving	10
parametric design (PD)	parametric design (PD)	8
Monte Carlo simulation	Monte Carlo simulation	6

of Appendix B illustrates the inner related topics in each emerging theme in the 2012–2017 time-slice.

• 2018–2023 Time-Slice

In the 2018–2023 time-slice (Figure 7d), there are 2225 published documents, distributed in the four typologies as follows:

(a)

Two thematic areas belong to motor themes (machine learning and building information modeling (BIM)). Their prominence reflects the growing shift toward data-driven and intelligent construction practices. BIM continues to evolve as a central platform for integrated project delivery, while machine learning enables advanced analytics, predictive modeling, and automation across various construction processes.

Table A17. Details for 2018–2023 motor themes.

Cluster Label	Related Topics	Occurrences
building information modeling (BIM)	building information modeling (BIM)	387
	industry foundation classes (IFCs)	66
	optimization	52
	digital twin	50
	virtual reality (VR)	48
	automation	46
	robotics	41
	internet of things (IoT)	40
	genetic algorithm (GA)	38
	blockchain technology	33
	modular construction	31
	ontology	30
	multi-objective optimization	27
	smart contracts	27
	simulation	25
	facilities management (FM)	24
	scheduling	24
	augmented reality (AR)	22
	semantic web	22
	architecture–engineering–construction	21
	generative design	20
	agent based modeling	19
	reinforcement learning	19
	finite element modeling	19
	construction engineering and management	18
	path planning	18
	geographic information system (GIS)	17
	parametric design (PD)	17
	interoperability	17
	planning	16
	construction automation	15
	linked data	15
	offsite construction	14
	inspection	14
	tower cranes	12
	maintenance	12
	prefabrication	12
	energy efficiency	12
	digitalization	11
	industry 4.0	11
	decision-making	11
	data integration	11
	decision support system (DSS)	11
	computational design	11
machine learning (ML)	machine learning (ML)	100
	point cloud	95
	unmanned aerial vehicle (UAV)	61
	structural health monitoring (SHM)	54
	construction safety	50
	artificial neural network (ANN)	35
	artificial intelligence (AI)	35
	construction	33
	terrestrial laser scanning (TLS)	31
	light detection and ranging (lidar)	27
	progress monitoring/tracking	25
	natural language processing (NLP)	25
	laser scanning	22
	long short-term memory (LSTM)	20
	construction machines	19
	construction site	17
	safety management	16
	random forest	15
	masonry	15
	generative adversarial network	14
	safety	14
	visualization	14
	3D modeling	14
	excavator	14
	quality control	14
	tunnel boring machine (TBM)	14
	pose estimation	14
	tunneling	14
	support vector machine (SVM)	13
	construction worker	13
	activity recognition	13
	productivity	13
	project management	12
	shield tunneling	12
	electroencephalography (EEG)	12
	scan-to-BIM	12
	data fusion	11
	photogrammetry	11
	earthmoving operations/projects	11
	structure-from-motion	11
	localization	11
	real-time monitoring	11

of Appendix B illustrates the inner related topics in each motor theme in the 2018–2023 time-slice.

(b)

Two thematic areas as emerging themes [3D printing and deep learning].

Table A18. Details for 2018–2023 emerging themes.

Cluster Label	Related Topics	Occurrences
3D printing	3D printing	50
	additive manufacturing	21
	digital fabrication	15
	reinforcement steel bars	13
deep learning (DL)	deep learning (DL)	231
	convolutional neural network (CNN)	99
	computer vision	91
	semantic segmentation	44
	object detection	35
	image processing	30
	ground penetrating radar (GPR)	29
	crack detection	26
	3D reconstruction	24
	non-destructive testing (NDT)	24
	transfer learning	15
	defect detection	14
	instance segmentation	13
	sewer	12
	segmentation	12
	concrete	12
	bridge inspection	11
	asphalt pavement	11
	attention mechanism	11

of Appendix B illustrates the inner related topics in each emerging theme in the 2018–2023 time-slice.

4. Discussion

4.1. Key Findings

This research provides a scientometric analysis and visualization for the development trajectories and trends associated with AICJ publications. This section affords the main results/findings that address the research questions.

• RQ1: How has AICJ’s research productivity and influence evolved from 2000 to 2023?

The research productivity and influence of AICJ from 2000 to 2023 can be summarized as follows.

• From January 2000 to December 2005, the number of publications is 322 articles while the total citations equal 538.
• From January 2006 to December 2011, the publication number is 536 articles, while the total citations equal 5492.
• From January 2012 to December 2017, the publication number is 1001 articles, while the total citations equal 25,257.
• From January 2018 to December 2023, the publication number is 2225 articles, while the total citations equal 111,091.

These trends indicate that every six years, the number of publications has increased by an average factor of 1.9, while total citations have grown by an average factor of 6.4. These factors reveal that the AICJ journal has become increasingly widespread and multidisciplinary over time, exerting a significant global influence on research in the construction and building technology field.

• RQ2: Which are the influential AICJ articles?

The most influential AICJ articles are identified based on the highest GCS and the DACY, as provided in Table 3. Notably, 19 of the top 20 publications based on GCS were published before 2015, whereas 12 of the top 20 publications based on DACY were published after 2015. These influential works extensively address and cover topics such as damage detection, segmentation, and classification; crack detection, segmentation, and classification; inspection of bridges, buildings, and roads; building information modeling (BIM); scan-to-BIM; digital twin; blockchain technology; laser scanning; additive manufacturing; robotics; safety; freeform construction; and energy performance.

• RQ3: What are the pioneering research entities published in AICJ in terms of counters, organizations, and authors?

According to the co-authorship study of countries, 70% (n = 56/80) are recognized in at least five AICJ documents with a minimum citation score (CS) of 50 or above. China (1179 documents, 40,556 citations) and the USA (923 documents, 43,240 citations) are the superior countries regarding the number of publications, total citations score, and cooperation intensity with 131 mutual documents. At the same time, Turkey, South Africa, Iraq, Denmark, and Vietnam are the top active countries with APY ≥ 2020, while Vietnam, North Ireland, Colombia, Luxembourg, and Australia have the highest ACD scores ≥ 50 citations per document.

According to the co-authorship study of the organizations, 7% of organizations (n = 149/2118) are recognized in at least ten AICJ documents with a minimum citation score (CS) of 100 or above. Hong Kong Polytechnic University is identified as the superior organization in terms of the number of publications and total citations score with 215 documents and 9436 citations. At the same time, Chinese Academy of Sciences, Hebei University of Technology, Nanjing Tech University, Hohai University, Teesside University, and Southwest Jiaotong University are the top active organizations with APY ≥ 2020, while University of Southern California, Heriot-Watt University, Georgia Institute of Technology, Hubei Engineering Research Center for Virtual, Safe and Automated Construction, and Curtin University have the highest ACD scores with ≥50 citations per document. In contrast, Hong Kong Polytechnic University-Tsinghua University (18 mutual documents) are identified as the top organizations regarding the cooperation intensity.

According to the authors’ co-authorship analysis, 1.1% of authors (n = 104 out of 9116) have authored at least ten AICJ publications, each having a minimum CS ≥ 100. Li Heng is noticed as the most productive and cited author, having published 72 documents with 4187 citations. Meanwhile, Jochen Teizer achieves the highest ACD score, averaging 136 citations per document. In contrast, Ankang Ji, Yue Pan, Limao Zhang, Vincent J. L. Gan, and Jiepeng Liu are the top active authors with APY ≥ 2021.

• RQ4: What is the conceptual knowledge structure of AICJ articles and how has it dynamically evolved from 2000 to 2023?

Concerning the science mapping, the co-occurrence analysis of AKs is used to classify AICJ publications into five scientific clusters. These clusters are associated with foundational optimization, decision-making, modeling and simulation applications (red cluster), deep learning and computer vision applications (green cluster), building information modeling applications and their interrelated topics (blue cluster), 3D printing applications (yellow cluster), and machine learning applications (mauve cluster). Furthermore, the AKs temporal evolution and distribution are evaluated over four successive time-slices as per Figure 7. The research front for these slices is investigated and classified into motor, basic, niche, and emerging themes based on importance (Callon centrality) and development (Callon density). It is worth noting that the themes from the 2018–2023 time-slices can be considered prevailing or dominant research areas. Its motor themes [building information modeling (BIM)] and machine learning] and their inner research directions shall be subjected to continuous development and exploitation since they are identified by their high density and high centrality. For the building information modeling (BIM) theme, this theme comprises building information modeling applications and their interrelated sub-research directions, including the following:

• Research Direction 1: [industry foundation classes (IFC), virtual reality (VR), augmented reality (AR), digital twin, internet of things (IoT), facilities management (FM)];
• Research Direction 2: [optimization, multi-objective optimization, genetic algorithm (GA), decision-making, decision support system (DSS)];
• Research Direction 3: [industry 4.0, digitalization, automation, robotics, path planning];
• Research Direction 4: [blockchain technology, smart contracts, geographic information system (GIS), interoperability, data integration, linked data, semantic web];
• Research Direction 5: [simulation, agent-based modeling, reinforcement learning];
• Research Direction 6: [computational design, generative design, parametric design (PD), finite element modeling];
• Research Direction 7: [modular construction, offsite construction, prefabrication, planning, scheduling, inspection, maintenance, energy efficiency, tower cranes].

For the machine learning theme, this theme comprises machine learning applications and their interrelated sub-research directions, including the following:

• Research Direction 1: [terrestrial laser scanning (TLS), laser scanning, light detection and ranging (lidar), point cloud, scan-to-BIM, data fusion];
• Research Direction 2: [structural health monitoring (SHM), masonry];
• Research Direction 3: [natural language processing (NLP), construction safety, safety management, construction site, quality control];
• Research Direction 4: [productivity, progress monitoring/tracking, real-time monitoring, unmanned aerial vehicle (UAV), photogrammetry];
• Research Direction 5: [tunneling, shield tunneling, tunnel boring machine (TBM)],
• Research Direction 6: [earthmoving operations/projects, excavator, construction machines, pose estimation];
• Research Direction 7: [construction worker, activity recognition, electroencephalography (EEG)].

In contrast, emerging themes [deep learning and 3D printing] shall warrant further investigation and exploration as they are characterized by their low density and low centrality. For the deep learning theme, this theme comprises deep learning and computer vision applications and their interrelated sub-research directions, including the following:

• Research Direction 1: [image processing, object detection, crack detection, defect detection, instance segmentation, semantic segmentation, 3D reconstruction];
• Research Direction 2: [non-destructive testing (NDT), ground penetrating radar (GPR), sewer, asphalt pavement, and bridge inspection].

For the 3D printing theme, this theme comprises 3D printing applications and their interrelated sub-research directions, including the following:

• Research Direction 1: [additive manufacturing, digital fabrication, reinforcement steel bars, concrete].

4.2. Contributions and Implications

In terms of theoretical contributions, this work is an endeavor to delve into, explore, and visualize the characteristics of AICJ literature. The advantages of the methodology implemented in this study in comparison to other review methodologies [7,8,9,10,11,51,52,53,54], can be realized as follows. First, regarding pioneering research entities, the findings are obtained based on not only NP but also on three additional metrics: APY, CS, and DACY. These metrics collectively reflect the productivity, activeness, influence, and document scientific value of such entities. Second, while previous studies typically used science mapping to identify publication clusters based on author-keywords, this work extends this approach by employing thematic evolution mapping. This provides a detailed view of the temporal development and distribution of publication clusters over time while offering a comprehensive understanding of the dynamic nature of AICJ research clusters and their time-to-time shifts. Moreover, it enables capturing of the dominant and promising research threads (motor and emerging themes of 2018–2023 time-slice) to act as a guidance for developing impactful and innovative studies aligned with AICJ scope and aim. Third, the reporting of analysis outcomes integrates both tabular and graphical patterns (such as information tables and network maps) with a narrative explanation to help the analysis results become clearer.

The research findings hold significant practical implications that can aid junior researchers, organizations, editorial boards, and practitioners in various aspects. Junior researchers can leverage the findings on authors’ cooperation patterns (Section 3.2.3), author-keyword clusters (Section 3.3.1), and the temporal evolution and distribution of author-keywords (Section 3.3.2) to identify collaboration opportunities with leading AICJ researchers and focus on dominant and promising research themes for further exploration and development. Universities can leverage the results on organizations’ cooperation patterns (Section 3.2.2), author-keyword clusters (Section 3.3.1), and the temporal evolution and distribution of author-keywords (Section 3.3.2) to establish academic partnerships with top AICJ publishing entities and apply substantial research themes to facilitate further investigation and analysis. Moreover, editorial boards may utilize the findings related to author-keywords clusters (Section 3.3.1) and the temporal evolution and distribution of author-keywords (Section 3.3.2) to adjust and refine the aims and scope of their sources’ publications. As for practitioners, they can apply the findings on authors and organizations’ cooperation patterns (Section 3.2.2 and Section 3.2.3) to engage with pioneering research experts for developing tailored solutions and fostering industry–academia partnerships with leading organizations.

4.3. Limitations

This study has a few limitations that should be stated. First, using data gathered and processed in 2024, this study examines AICJ publications from 2020 to 2023. In order to provide more comprehensive benchmarking and contextual insights, future research is encouraged to periodically update the analysis with the most recent data and expand the comparison to include top construction journals. Second, the study is limited to peer-reviewed English articles that are obtained from the WoS database. Third, the analyses’ outcomes for VOSviewer and Biblioshiny may differ slightly depending on the parameter settings used. Fourth, without examining the articles’ core contents, this research crystallizes the primary clusters and frontiers of AICJ research based on the AKs’ co-occurrence and their temporal evolution and distribution. Finally, the focus of this study is limited to AICJ. However, this methodological decision is intentionally made due to the journal’s significant role and high impact in the construction field, its trendsetting influence, and its wide-ranging multidisciplinary scope coverage. This targeted approach enables a thorough understanding of innovative developments and academic discussions within a key source of construction technology. Nonetheless, comparing AICJ with other leading construction or civil engineering journals could provide better context for the findings. This would provide a broader perspective on AICJ’s unique role and contributions within the overall academic field. In addition, this work can be expanded upon in future work through content analysis in conjunction with a scientometric analysis in order to provide additional insights and broaden the findings presented here.

5. Conclusions

This research analyzes the AICJ literature by conducting a scientometric analysis, which is implemented through a three-stage approach. First, using the WoS Core Collection database, 4084 publications between 2000 and 2023 are identified and obtained for analysis. Second, the publications are analyzed through co-authorship analysis, keyword co-occurrence analysis, and keyword thematic evolution, along with VOSviewer and Biblioshiny as software tools. Third, the analysis is carried out through the following sequential steps: (1) assessing the output of the publication and citation and identifying the influential research works; (2) looking into the networks of researchers, countries, and organizations that collaborate in AICJ literature; and (3) exploring and analyzing the conceptual knowledge structure of AICJ literature using author-keywords.

The study’s findings show a notable rise in AICJ’s research output and scholarly influence, particularly in recent years (2018–2023), accounting for 54.5% of total publications and 78.0% of total citations, highlighting the journal’s growing prominence. Moreover, the findings indicate that 70% of the countries have contributed to more than five AICJ documents, with a minimum CS of 50 or higher; 7% of organizations have published at least ten influential AICJ documents, with a minimum CS of 100 or higher; and 1.1% of authors are acknowledged in at least ten AICAICJ documents with a minimum CS of 100 or higher. Furthermore, the AICJ literature was classified into five major research clusters: (a) foundational optimization, decision-making, modeling, and simulation applications; (b) deep learning and computer vision applications; (c) building information modeling applications; (d) 3D printing and robotics applications; and (e) machine learning applications. This recent period (2018–2023) also aligns with significant advancements in construction research, notably characterized by accelerated progress in digitalization, primarily through building information modeling (BIM) and computational design methods. Additionally, the rapid adoption of artificial intelligence (AI) and machine learning techniques, aimed at enhancing automation and predictive analytics, was evident. Finally, increased utilization of robotics and 3D printing technologies has been driving more sustainable and innovative construction practices. These findings can assist readers and researchers in gaining a comprehensive understanding of AICJ published work, help research groups plan and optimize their efforts, and guide editorial boards to focus on the most promising areas in the current body of knowledge for future research.

Accordingly, future AICJ research could be directed at further developing and exploiting motor themes (building information modeling (BIM) and machine learning), which are recognized by their high density and centrality. For BIM, key directions involve digital twinning, decision support systems, industry 4.0 applications, blockchain technology, reinforcement learning, computational/generative design, and modular integrated construction. For machine learning, key directions comprise point cloud processing, structural health monitoring, natural language processing using large language models and Retrieval Augmented Generation (RAG), real-time tracking and monitoring, tunneling/TBM-related applications, and worker/equipment activity recognition. In contrast, emerging themes (deep learning and 3D printing) could be subjected to further exploration, particularly in areas related to object vision-based applications, non-destructive testing (NDT), additive manufacturing, and digital fabrication.