Análisis bibliométrico mundial de BIM a través de la colección principal de la Web of Science (2003-2017) ; Global bibliometric analysis of Building Information Modeling through the Web of Science Core Collection (2003-2017)

The main objective is to perform a quantitative and qualitative analysis from the institutional point of view of the related research on Building Information Modeling (BIM) collected in the Web of Science Core Collection during the period 20032017. For this, all documents of the type ‘article’ or ‘review’ were reviewed and several bibliometric indicators analyzed. Similarly, the international dissemination of the research of the 16 most productive countries was analyzed, indicating the publication journals of each country and concluding that the main journals. USA is the most productive country although with a relative impact factor slightly higher than average due to the dispersion of its publications in journals of medium or low impact. It also shows how Curtin University (Australia) and the Georgia Institute of Technology (USA) have been the most productive and high impact institutions, although the outstanding rise of Hong Kong Polytechnic University is noted.


INTRODUCTION
As one of the most influential innovations in the construction industry, Building Information Modeling (BIM) is able to facilitate work in different areas such as construction, industry, manufacturing and facilities management (1)(2)(3)(4). The concept of BIM was suggested by Eastman (5), and later used by Van Nederveen and Tolman (6) and Tolman (7), and can be described as an integrating technology that modifies the digital process of building representation (8).
Given the great thrust of publications related to BIM in its different areas or categories, it has been considered appropriate to carry out a bibliometric analysis different from those carried out to date, in order to extract results from an institutional point of view at both the country and research center levels of research, using the Web of Science (WoS) Core Collection database as the main worldwide database given that the main scientific journals are identified with Impact Factor (IF), and with a minimum quality level governed by the pairs review system (blind or double blind peer review).
Institutional studies of this type, where scientific publications are analyzed through quantitative and qualitative indicators, have been carried out previously (9)(10) (11). The qualitative analysis can be carried out through various indicators such as the number of citations received (12), through the h-index (13), the Eigenfactor (14), or through the IF (15), among others. However, this last method is the most widely accepted by the scientific community, despite the criticisms of its use (16) (17), since for its calculation only the last 2 years are chosen, when the tendency to change is much slower, or that a document is considered to be of higher quality the more times it is cited.
As a first approximation in the field of construction, it is convenient to indicate that there are interesting publications where bibliometric analyses of some main journals of the Journal Citation Reports (JCR) (18) (19) are carried out, or of the Spanish scientific publications in the Category 'Construction & Building Technology of the JCR (10), although not specifically BIM.
After an exhaustive revision in WoS of the publications that deal with bibliometric analysis related to BIM, some of them have been found to be very interesting. From a bibliometric review of improvements in building maintenance (20), research on BIM in China (21), or the analysis of Brazilian scientific production on BIM in the period 2000-2015 (22), passing through bibliometric analysis to identify research clusters (knowledge domains or knowledge bases) and topics in the BIM community (23), studies through co-author analysis, co-word analysis and co-citation analysis (24), bibliometric analysis to characterize the 8 emerging categories of work related with BIM as well as the most researched topics (25), studies on collaboration in BIM-based construction networks (26), analytics for mapping the managerial areas of Building Information Modeling (27) and finally conducting studies to analyze the evolution in the intellectual structure of BIM research (28).
However, there are no studies worldwide that use other quantitative and especially qualitative bibliometric indicators such as the Impact Factor, h-index or the average number of citations per document to establish ranking among countries and research centers (29) (30). There is as well no detailed analysis of the dissemination of research related to BIM, in the main international journals and for the most productive countries. Similarly, the detail of the collaboration network between countries and especially between research centers has been deepened in order to detect the three main collaborators and the percentage of their collaboration.
For all these reasons this original research, based on the largest number of records analyzed to date, is justified so that at an institutional level both countries and research centers can develop strategies within the framework of scientific technology policy and seek synergies with other institutions in order to increase the visibility and impact of their publications regarding BIM.

Source of information for the extraction of scientific production
The database selected for the analysis of the scientific production on BIM was the Web of Science (WoS) 'Core Collection' of Clarivate Analytics, because although there are other alternative databases such as Scopus, Compendex or Inspec, WoS performs the Impact Factor study by analyzing its visibility, so it has been necessary to download all the impact factors of all journals for the entire time series (2003-2017).

Extraction process
In order to download all the records to be analyzed the WoS  [28], and Art Exhibit Review [1]).
The typologies found were Book, Journal and Series, and given that a qualitative analysis based on the Impact Factor is the aim of this study, only type Journal was selected. Subsequently a refinement was applied by type document (Article or Review) (documents from here on), resulting in a total of 1,332 documents [1,244 articles, 28 article proceedings, and 60 reviews] published in 263 journals by authors from 69 countries.

Construction of 'ad hoc' database for analysis
The download made it possible to save the fields indicated in the previous section for each record, making it possible to build an 'ad hoc' database with Microsoft Access 2016, which has allowed consultations to be carried out in a simple, flexible, and fast way for the different bibliometric indicators.
Thus each document is awarded to each of the countries or research centers, allowing multiple counting, provided that said country or research center appears in the Research Address field of the database.
to all. However, it is a first approximation that can be debated. Thus, in order to obtain IFw from a country or research center, the Total Impact Factor (TIF) of that country or research center is divided by the number of documents from that country or research center.
Similarly, in order to obtain IFr from a country or research center the IFw of that country or research center is divided by the IFw of the series analyzed (in the case of countries, the series is constituted by the 33 most productive countries, and in the case of the research centers the 54 most productive), with which we can determine the relative position of each of the countries or research centers in these series. Thus, if the IFr is higher than the unit that country or research center will be placed above the value of the series, and vice versa, if it is lower than the unit, it will be positioned below.
Finally, the dissemination and internationalization of journals has been analyzed as a starting point for the qualitative analysis of scientific production.

BIM-related documents
Once the data had been downloaded, refined and extracted, the world scientific production on BIM was classified by document type.
From Figure 1 it can be seen that from the year 2003, in which the first document on BIM indexed in WoS appears, until 2017, the evolution of the number of documents shows a slight slope of growth, but from the year 2012 is when it begins to grow remarkably, and the same happens with the authors and with the research centers that present a slope of greater growth from the year 2009. Finally, it should be noted that the citations showed a remarkable growth from 2007 to 2013, the date to from which they are decreasing as they approach the current date, and that national and international collaboration between countries also shows a slightly higher growth in national collaboration than in international collaboration.
From Figure 2, it can be observed that the average number of authors and research centers per document has always been oscillating in the time series, as well as the average number of citations per document (reaching a maximum in 2009), but with a clear downward trend since articles with more citations tend to be the oldest of the time series. The IFw has also been oscillating, although there is a clear growing trend since 2014. Even so, and as has been verified, there are problems of normalization in WoS, since the same research center found different denominations, which forced us to check each of the records one by one in order to construct a base of refined and reliable data. In addition, some records were also found with the Research Address field empty, so they were eliminated from the bibliometric analysis.

Quantitative analysis
The bibliometric indicators analyzed from the quantitative point of view were: Therefore, a document signed by authors from different research centers will count equally in each country or research center.
A similar study has been carried out for each document of the author keywords, keywords plus, and the language of publication.

Qualitative analysis
The bibliometric indicators analyzed from the qualitative point of view were: Cita: Average number of citations per document assigned to a country or research center. h-index: Hirsch index assigned to each country or research center.

IFw & IFr:
Weighted IF and Relative IF per document corresponding to a country or research center. First, and before calculating IFw and IFr, the Total Impact Factor (TIF) of a country or research center must be calculated.
To calculate TIF, the Impact Factor of the journal is assigned to each document of the country or research center in the year of publication of the document, and so on for each and every one of the documents of the country or center of research. These Impact Factors are then added up and the TIF of the country or of the research center calculated.
However, this methodological procedure for assigning qualitative indicators to a country or research center is not standardized, since as is well known the IF is a value that belongs to a journal and not a document. In a similar manner, the documents published in the same year as the IF do not relate to the articles selected to calculate the IF (two years ago), and on the other hand, in the same journal there are documents that have been cited many times and others fewer, for which reason it would not be exact to award the same citation value versity Bulletin), which underlines the residual value of the other languages.

Trends in research
Research trends have been observed after analyzing the main keywords (author keywords and keywords plus). Thus, the 36 author keywords and 46 keywords plus most used for the time series and for each of the 4 sub-periods studied (2003-2006, 2007-2010, 2011-2014, and 2015-2017) were analyzed. The practical absence of information in the first sub-period (2003)(2004)(2005)(2006) is to be ruled out as shown in Figure 1, in those four years only 11 documents were published. total], evidencing the opportunity of writing in that language in order to seek wide dissemination and visibility (31). They have also been published in other languages such as German [26]

Research activity according to country
An analysis for the 33 countries with the highest scientific output indicating the number of documents, the percentage of documents in national and international collaboration, the average impact factor per document, the average number of citations per document, the average number of authors per document, the average number of research centers per document and the h-index has been carried out (Table 1). Within the keywords plus with more appearances four stand out for their tendency in the ranking: System, Design, Management and Model, which have almost always remained among the top 5 positions. Other keywords plus that stand out for the growth of their ranking between the last 2 sub-periods are: Innovation, Adoption, Ontology, and Life-Cycle. On the contrary, the keywords that lose importance are CAD, 3D, and Energy. Therefore, the author keyword and keyword plus most commonly used in both has been Ontology.  [28], the Technion Israel-Institute of Technology [26] and Yonsei University [25], assuming the assumption of 4.13%, 4.05%, 2.10%, 1.95% and 1.88% respectively of the total scientific production.
With regard to collaboration with researchers from the same research center, it has become clear that within the 5 most productive research centers the It has also been possible to analyze the average number of authors per document, underlining the fact that almost all countries have values that range between 3 and 5, highlighting Sweden as the country with the lowest value of the whole series [2.31].
Finally, the h-index is analyzed in which the USA stands out with a value of 37 among the five most productive countries. This shows the quality of the publications of that country, presenting such a large number of the articles cited. However, this result must be qualified because the number of US documents [329] is much higher than the rest. Thus, if the ratio of this indicator is set against the number of documents the USA and the Peoples Republic of China would occupy the last places in this indicator, highlighting countries such as Ireland, Scotland, Slovenia and Portugal, although as mentioned before it is much more difficult to maintain the same ratio as these countries when the number of documents is very high.
Also, the scientific production disaggregated by sub-periods has been studied. We can see how the 5 most productive countries are generally maintained within the first 5 positions, with the USA always in first place. On the other hand there are other countries that are rising in the ranking notably like Italy, Malaysia, New Zealand and Singapore, if the last two sub-periods are examined, and on the contrary other countries like Japan, Scotland, Portugal and Slovenia have descended in their positions considerably.

Research activity according to research center
The search for research centers in the Research Address field of the ad hoc built database has demanded an immense effort. Each of them has had to be revised because the standardization errors existing in WoS need to be taken into account, since a good number of research centers presented different denominations. Also, the 54 research centers ordered by On the other hand, from the analysis of collaboration between the 38 research centers with the highest scientific output it can be said that by analyzing the first 3 collaborators of the 5 most productive research centers, we observe how there is an appreciable collaboration between 2 of them (the Georgia Institute of Technology and the Technion Israel-Institute of Technology), a low collaboration ratio of Hong Kong Polytechnic University and also Yonsei University, and no relationship of Curtin University with any of the remaining 4. The latter generally presents a low international collaboration, and practically does not collaborate with any of the 15 most productive research centers, which demonstrates a high rate of collaboration of researchers from the same center or from different research centers in the same country of origin (Australia).
Also noteworthy is the collaboration of some research centers which, although they are not among the 10 most productive, do present percentages of collaboration over 25% with their first country of collaboration for a number of documents. The four research centers in Hong Kong always have among their three main collaborators one of these institutions, as well as the three Chinese universities which also collaborate with them.

Internationalization and dissemination of journals
Interesting observations can be extracted in

National and international collaboration networks between countries and research centers
After analyzing national and international collaboration, it has been possible to search the database to determine the international relationship between the 3 main collaborators by number of collaborations, establishing the percentages through the number of documents collaborated upon.
In the same way, the national and international relationship between research centers was defined, and therefore the collaboration networks at national and international level. Also, the international collaboration network among the 33 countries with the highest scientific output has been analyzed.
From the analysis of international collaboration between countries it is first of all worth noting that the USA is the first country with the most productive international collaboration, highlighting South Korea with 72.92% of its scientific production, Canada with 64.71%, Turkey with 63.64 % and Israel with 60%. Similarly, and in the opposite direction, the US collaboration with South Korea represents 29.41% of its scientific production, with the Peoples Republic of China 28.57%, and with Germany 11.76%. except one (Austria): Automation in Construction. This becomes, therefore, the main journal in which issues are published related to BIM worldwide. Furthermore, it has an impact factor during all the years of the time series.
On the other hand, it is striking that there are journals that bring together 100% of their scientific production with authors from the same country. This is the case of Agro Food Industry Hi-Tech with the Peoples Republic of China, Bautechnik with Germany, the Journal of Asian Architecture and Building Engineering with South Korea, and Jurnal Teknologi with Malayia. This can favor inbreeding at the 16 most productive countries, among all the documents published by each journal, and in parentheses the percentage of all the scientific production of each country in the journals, as long as it is equal to or greater than 10% of the total scientific production of the country in the entire time series (2003)(2004)(2005)(2006)(2007)(2008)(2009)(2010)(2011)(2012)(2013)(2014)(2015)(2016)(2017). Therefore, given that in a document there may be authors from different countries, this document will be computed equally in each one of them. Of these, 19 have an impact factor for the entire time series (2003)(2004)(2005)(2006)(2007)(2008)(2009)(2010)(2011)(2012)(2013)(2014)(2015)(2016)(2017).
From this analysis it is deducted that only one journal presents publications from the 16 most productive countries (The values in parentheses correspond to the percentage of production for each country of its total scientific production).

CONCLUSIONS
In this article an in-depth bibliometric analysis of all the BIM-related records present in WoS Core Collection database in the 2003-2017 time series has been carried out. A total of 1,332 records of the article and review typology (documents) were analyzed in order to obtain a series of quantitative and qualitative bibliometric indicators which would help to obtain an overview of everything published in this period regarding BIM. The information obtained in this research is very useful from the institutional point of view, since it will help countries and research centers to develop strategies to strengthen their scientific policies and to increase the visibility of their research, establishing a ranking among countries and research centers with higher scientific production.
From the quantitative point of view, several indicators have been obtained for each year of the time series such as the number of documents, citations, authors, research centers, the number of documents in collaboration between authors from the same country (national collaboration), the number of documents in collaboration between authors from different countries (international collaboration), the number of documents between researchers of the same research center and the number of documents between authors from different research centers. Similarly, the most frequent author keywords and keywords plus have been obtained for each of the 4 sub-periods into which the time series was divided, which indicates the trends in research in the time series. On the other hand, it is confirmed that journals with a low impact factor publish articles from countries with lower scientific production, presenting a high percentage of articles that come from a single research center and with much less international diffusion.
An analysis of the dissemination of research has also been carried out through the 25 journals with the highest scientific output for each of the 16 most productive countries. For this, we have obtained, on the one hand, the percentage of articles from each country in each journal, and on the other hand, the percentage of scientific production of each country in each journal. As a main conclusion, it has been found that only the journal Automation in Construction presents publications of all that countries.
research center the average Impact Factor, the relative Impact Factor and the h-index have been obtained.
Analyzing the quantitative results it can be observed that the growth trend of the theme is smooth, considering that in the 15 years of the time series only 1,332 documents of the type article or review were published. These are the documents from which the Impact Factor is obtained, which means an average of only 89 documents per year. However, a notable growth is observed from the year 2012, also for the number of authors and research centers.
The documents we found had been published in 263 journals, with English being the main communication language. Of these the journal Automation in Construction brings together almost 21% of all scientific production, highlighting others such as Advanced Engineering Informatics, the Journal of Computing in Civil Engineering, and the Journal of Construction Engineering and Management.
On the other hand, research trends have been studied through the analysis of author keywords and most frequent keywords plus. Thus, it has been observed that among the author keywords four stand out for their position in the ranking, such as Industry Foundation Classes (IFC), Information Technologies, Interoperability and Facilities Management.
Within a geographical and institutional context 69 countries have participated in all the scientific production unloaded, among which the USA stands out with 329 documents, followed by the Peoples Republic of China, England, South Korea, and Australia, highlighting South Korea in national collaboration, and the Peoples Republic of China and Australia in international collaboration.
The qualitative assessment of these countries has been carried out through the analysis of the relative Impactor Factor (IFr), and of the h-index. Thus, among the 5 most productive countries England stands out with the highest IFr, the USA with the highest h-index, and Australia as the country with the best relationship between the h-index and the number of documents. In addition, it can be observed that almost all the G8 countries (USA, Canada, Germany, Italy, France, England, Japan and Russia), except Japan and Russia, are