Digital Transformation Strategies, Practices, and Trends: A Large-Scale Retrospective Study Based on Machine Learning

Abstract

The purpose of this research is to identify the areas of interest, research topics, and application areas that reflect the research nature of digital transformation (DT), as well as the strategies, practices, and trends of DT. To accomplish this, the Latent Dirichlet allocation algorithm, a probabilistic topic modeling technique, was applied to 5350 peer-reviewed journal articles on DT published in the last ten years, from 2013 to 2022. The analysis resulted in the discovery of 34 topics. These topics were classified, and a systematic taxonomy for DT was presented, including four sub-categories: implementation, technology, process, and human. As a result of time-based trend analysis, “Sustainable Energy”, “DT in Health”, “E-Government”, “DT in Education”, and “Supply Chain” emerged as top topics with an increasing trend. Our findings indicate that research interests are focused on specific applications of digital transformation in industrial and public settings. Based on our findings, we anticipate that the next phase of DT research and practice will concentrate on specific DT applications in government, health, education, and economics. “Sustainable Energy” and “Supply Chain” have been identified as the most prominent topics in current DT processes and applications. This study can help researchers and practitioners in the field by providing insights and implications about the evolution and applications of DT. Our findings are intended to serve as a guide for DT in understanding current research gaps and potential future research topics.

1. Introduction

In today’s world, rapidly developing digital technologies and increasingly adopted innovative and competitive business models have compelled many businesses to undergo digital transformation (DT) in recent years. DT refers to businesses of all sizes leveraging online resources and technology to improve the efficiency of their products, services, customer experiences, workflows, and decision-making processes [1,2]. Virtual reality, digital literacy, the Internet of things, digital employees, big data, machine learning, cloud services, and blockchain have increased the adoption of DT, allowing organizations to improve service quality, efficiency, profitability, and customer satisfaction [1,3]. With the increase in online resources and the acceleration of technological developments, DT has become a strategic process that closely concerns businesses [2].

This process has resulted in radical shifts in the strategic decision-making paradigms, business models, and services of businesses in every country and industry. Emerging digital technologies such as artificial intelligence, robotics, big data analytics, the Internet of things, and blockchain play an important role in increasing company efficiency and lowering costs by enabling data-driven decision-making processes [4]. Furthermore, the problems encountered during the COVID-19 pandemic process clearly demonstrated the need for DT in many public and industrial areas. This case has accelerated the process of organizational DT. According to Statista, spending on DT was USD 1.18 trillion in 2019, with a projected increase to USD 1.8 trillion by 2022. This figure is expected to rise to USD 2.8 trillion by 2025 [5]. To be successful in global competition, businesses must adapt quickly to competitive business strategies and be constantly renewed. The most significant of these innovations is DT. The strategic significance of DT in today’s global competitive environment is growing by the day [6].

Academic research on DT is becoming increasingly important as the importance of DT for businesses and societies grows. DT research has recently increased as a result of the increased interest in this field [4]. Academic research on DT covers a wide range of research topics and draws important conclusions about business success in DT processes and global competition [3]. Academic research on the applications and effects of DT in various fields includes efforts to investigate how this transformation affects social, cultural, economic, and technological developments. Furthermore, numerous studies on DT processes and applications have been conducted in the context of various countries, governments, regions, organizations, or industries [7,8]. While the increase in DT studies provides important inferences and insights into the field, the exponential increase makes understanding the themes and focal points of the field-specific research landscape from a holistic perspective difficult [1]. Understanding DT research trends, themes, and application areas will make it easier for new researchers to recognize this field and identify potential research gaps [4]. Understanding these trends, issues, and changes is also important for developing a vision for the future of DT. Scientific literature is a rich and important source of information in this context. Using automated text mining procedures based on machine learning is one of the most effective ways to analyze the scientific literature [9,10]. Because automated text mining efficiently summarizes a large number of documents, it enables the discovery of hidden semantic topics in that collection [10,11,12].

Studies based on a DT literature analysis, on the other hand, are mostly of the systematic review and bibliometric analysis variety, with a relatively small number of studies. These studies primarily concentrated on the DT of various industries, including education [13], construction [14], and manufacturing [15]. It has been observed that studies dealing with DT in general have a small number of articles, such as systematic literature reviews [16] or bibliometric analysis [17]. Automated text mining research focuses on the temporal effects of DT [18] and its industrial applications [19].

Despite these commendable efforts, no study based on automated text mining has addressed DT as a whole. This study’s goal is to identify the research and implementation topics and their temporal trends by analyzing research on DT over the last decade using automated text mining and topic modeling procedures. The field of DT has been approached from a broad perspective for this purpose, and themes and trends have been identified using a machine learning-based topic modeling algorithm. A systematic taxonomy map for DT associated with the obtained topics has been created. Furthermore, the discovered topics’ trends and temporal developments are revealed. Because current reviews for DT have focused on a small number of articles or on specific industries, it is expected that this study will contribute to the literature with more specific and deep insights than the others. Answers to the following research questions (RQs) will be sought in the context of evaluating the previous decade of DT.

RQ1: What are the bibliometric characteristics of DT research?

RQ2: What are the main topics of DT research?

RQ3: How do DT topics fit into a taxonomy?

RQ4: What are the temporal trends in DT research topics?

2. Background

The concept of DT, its sub-contexts, and industrial application areas will be discussed in this section. The topic modeling method we used in our analysis, as well as its applications, will then be discussed.

2.1. Digital Transformation

The concept of digital transformation first appeared in the literature in the late 1990s and has since become a popular research topic due to the COVID-19 pandemic [20]. There are numerous definitions of DT in the literature, but no agreement has been reached on any of them. DT is defined as “an essential process of change driven by the innovative use of digital technologies, combined with the strategic leverage of key resources and capabilities, aimed at radically improving an organization and redefining the value proposition for its stakeholders” [21]. This process revolves around four basic elements: the unit where the DT will occur, the scope of the transformation, the tools, and the transformation’s outcome [3].

Information and communication technologies (ICT), which allow for the rapid acquisition, storage, processing, and distribution of information, are becoming more prevalent, increasing their importance in social and industrial fields [22]. This rapid development of ICT has provided various opportunities for organizations to innovate, but it has also necessitated this renewal. Organizations must adapt their business processes to SMACIT (social, mobile, analytics, cloud, and Internet of things) technologies in order to capitalize on emerging opportunities and obligations. This has resulted in the emergence of smarter and more effective new business models as a result of the DT of many small and medium-sized businesses (SMEs) [23]. In today’s global economy, digital transformation processes and practices provide significant benefits for SMEs to maintain competitiveness, efficiency, and profitability. A successful DT for organizations is only possible when technology is backed up by process and people [24].

Technology in the context of DT can be defined as a tool that assists people in moving forward more efficiently by automating work. When it comes to DT, technology is usually the first thing that comes to mind. Despite all of its benefits, thinking of DT solely in terms of technology can lead to organizations making serious mistakes and failing their DT initiatives [25]. At this point, the process component enters the picture to assist the technology component. A process is a collection of actions that must be carried out in order to achieve a goal [26]. Processes, on the other hand, are complex activities that must be completed during the transformation phase [27] because a process should be designed in such a way that it can meet the needs of the organization’s stakeholders while also generating values that contribute to the organization’s operational excellence. In process design, it is necessary to establish a collaborative DT vision in the organization, understand the organization’s business model and processes, and evaluate the organization’s DT performance and capabilities. Furthermore, reviewing organizational policies, taking into account organizational culture, and developing and testing processes [28] are also necessary. Aside from the other two components, one of the most important aspects of DT is the human component. Internal stakeholders are also directly linked to an organization’s DT. According to one study, two of the most important reasons for a failed DT process are related to the human component. The first is due to a lack of managerial support, while the second is due to a lack of skills in developing and implementing the organization’s digital strategy [27,29]. As a result, for an effective DT, managers must focus on their employees and prepare them for this transformation in terms of skills and mindset. This preliminary planning is critical not only for the success of the DT process, but also for the long-term viability of the DT activity.

A number of studies have been conducted to examine the DT literature in various contexts. In terms of scope, these studies can be divided into two categories. The first category includes studies that evaluate DT in general, while the second includes studies that focus on specific industries. When the literature is examined, it is discovered that systematic literature review studies on DT were not conducted until 2014 [30]. Henriette et al. examined 13 articles and conducted a managerial literature review on DT [31]. Reis et al. conducted a systematic review of the DT literature, reviewing 206 articles [4]. Gebayew et al. examined DT in the context of research by reviewing 30 articles [30]. Teichert examined 24 studies on DT maturity in 2019 and revealed the characteristics of 22 digital maturity models [32]. Hanelt et al. examined 279 studies in order to assess DT from the standpoint of organizational differentiation [16]. Zhu et al., on the other hand, used bibliometric analysis on 865 studies to provide an overview of DT [8]. Verhoef et al. conducted a scoping study on 84 studies in order to identify the three stages of DT [6]. Researchers conducted a systematic literature review in 2022 to evaluate Latin America’s DT [33]. In a similar vein, Chawla and Goyal conducted a bibliometric analysis on 234 studies to provide an overview of DT [23]. Facin et al. also conducted a bibliometric analysis on 294 studies [17].

On the other hand, DT is a fascinating topic for researchers from various disciplines. As a result, some studies in the literature focus on the applications of DT in various industries. Martin-Pena (2018), for example, investigated DT business models in the manufacturing industry and reviewed 65 studies to better understand the relationship between service and digitalization [15]. Cortellazzo, Bruni, and Zampieri (2019) examined 54 leadership-related DT studies [34]. Benavides et al. examined 19 studies to identify the characteristics that distinguish the DT processes in higher education from one another [13]. Olanipekun and Sutrisna (2021) examined 151 studies to determine the effects and challenges of DT applications in the construction industry [14]. The tourism industry is another area in which DT focuses. A bibliometric analysis of 535 studies was performed in order to better understand the DT in cultural heritage management [35].

2.2. Topic Modeling

Systematic literature studies using manual methods necessitate more labour and time. Bibliometric studies offer the benefit of a quick examination of the subject as well as a general evaluation. However, it does not provide detailed information about the content [36]. As a result, automated methodologies based on machine learning, such as topic modeling, enable effective analysis that can provide much deeper insights.

Topic modeling is a technique for detecting hidden semantic structures known as “topics” in large, unstructured document collections [10,11]. Latent Dirichlet allocation (LDA) is a three-level hierarchical Bayesian model-based generative probabilistic topic modeling approach [37]. In topic modeling research, the LDA model is frequently used [38]. LDA is based on the notion that each document is likely to contain one or more topics, and that collections of words that reflect a topic are used [11]. The topic distribution for each document and the word distribution for each topic are calculated using iterations of the Dirichlet distribution in this algorithm [37].

Topic modeling is also widely used in social science research due to the methodological advantages it provides. Gurcan et al. (2021) analyzed 41,720 articles using topic modeling to determine the trends and knowledge-domains of human–computer interaction [39]. Another case study conducted by Aziz et al. (2022) used topic modeling analysis on 5942 articles to search the finance literature [40]. However, there are only a few studies in the field of DT that use the topic modeling approach. Lopez and Wilfredo (2022) used a topic modeling approach to analyze 700 articles in order to better understand DT in a crisis situation [18]. Lee et al. used topic modeling and bibliometric analysis on 99 studies to identify strategic boundaries in advanced manufacturing and engineering and to understand DT trends [19]. Similarly, Chen et al. used topic modeling and bibliometric analysis to identify the main components of effective academic–industrial collaboration topics and patterns [41]. To the best of our knowledge, our study is the first to use a topic modeling approach to examine DT from a holistic perspective. As a result, it is expected that this research will contribute to future DT research.

3. Materials and Methods

This study’s methodology consisted of five sequential stages (see Figure 1) in order to investigate up-to-date strategies, practices, and trends of DT. First, a search strategy was developed in accordance with the theoretical background of DT in order to obtain the most consistent data within the scope of the study. With the articles obtained through this search, an empirical corpus for DT was then compiled. Applying preprocessing tasks to this textual corpus made it suitable for topic modeling analysis in the subsequent step. At this stage, the corpus was analyzed using a topic modeling technique based on Latent Dirichlet allocation (LDA). Finally, the content and background of the topics generated by LDA were interpreted, the topics were labeled, and the time-based accelerations of the topics were revealed. These procedures are illustrated in Figure 1.

3.1. Search Strategy and Data Collection

One of the most important steps in topic modeling research is the creation of the appropriate corpus; because the empirical corpus to be created is directly related to the performance of the results, the data collection should be done methodically [10,36]. In content analysis studies on scientific articles, various bibliometric databases can be used. Scopus, published by Elsevier, is one of the most widely used bibliographic databases [42]. With its large number of journals and advanced search filters, this bibliometric database provides access to more articles. As a result, Elsevier’s the Scopus bibliographic database is preferred in this study. This bibliometric database indexes a wide range of publication sources, including book reviews, editorial materials, book chapters, conference proceedings, and peer-reviewed journal articles. Peer-reviewed journal articles, on the other hand, consistently reflect the research environments of the disciplines. To demonstrate an objective approach in the relevant study, only peer-reviewed journal articles were included in the experimental data set in this context. In this context, a Scopus database query was made with “digital transform*” in the title, abstract, and keywords to create an experimental data set for DT. The desired articles were extracted from the Scopus database as a result of this search query, and an experimental data set was created that included bibliometric indicators, author keywords, and abstract and title sections of each article [11]. As a result, the experimental corpus of DT created includes 5350 peer-reviewed journal articles (4927 research articles and 423 reviews) published between 2013 and 2022.

3.2. Data Preprocessing

Data preprocessing is a fundamental data mining step that includes optimization processes to clean up noisy data on the raw data set prior to the construction of a model. This step is critical for text mining and natural language processing studies because it has a large impact on the performance of the analysis results [9]. To filter out noisy information from the extracted DT experimental corpus, we perform data preprocessing in five steps. To begin, each word was converted to lowercase. Tags, web links, punctuation, publisher information, symbols, and numeric expressions were then removed from the dataset. After that, tags, web links, punctuation, publisher information, symbols, and numeric expressions were removed from the dataset. All English stop words (how, what, and, is, or, the, a, an, for, etc.) were deleted in the fourth stage [11,43]. Finally, the remaining words were subjected to the lemmatization process, which resulted in the reduction of the words from their derived form to their plain form [44]. Following data preprocessing, each article in the dataset was converted into a word vector using the “bag of words” method to provide a numerical representation of the words in the corpus [10]. By combining these vectors, a document-term matrix in numerical matrix format representing the entire corpus and required for topic modeling analysis was created [45].

3.3. Implementation of the LDA-Based Topic Modeling

Topic modeling is a probabilistic approach based on machine learning that is used to manage, understand, and summarize large collections of text information [10]. It enables the detection of hidden dependencies between various patterns associated with specific topics in textual document sets. As a result, it is possible to identify word groups that best represent the information embedded in documents known as topics [38]. We use experimental compilation topic modeling with DT articles in this article. Our goal is to generate abstractions of semantic mappings of articles stored in topic sets. To obtain the patterns and topics included in the DT research corpus, we use the Latent Dirichlet allocation (LDA) algorithm. The LDA topic model is a probabilistic approach. LDA topics are less likely to have semantic overflow from documents and are more interpretable than topics produced by other algorithms such as Latent Semantic Indexing (LSI) or Probabilistic Latent Semantic Analysis (PLSA) [11]. LDA does not require a predefined tag or training set because it is based on an unsupervised machine learning model [45].

We used the tmtoolkit package [43], which includes the LDA model for topic modeling, to apply LDA-based topic modeling to the experimental corpus of DT research. Given our dataset’s preprocessed articles as input, the LDA algorithm generates a topic list that groups the articles under K topic labels. To determine the optimal number of K topics, we used the C_V consistency score, a standard practice recommended in previous studies. First, we ran the LDA model on our dataset with 1000 iterations, increasing the number of K topics by one each time from 5 to 60. We calculated the C_V consistency score of the generated topic model for each K at the same time [46]. A higher consistency score indicates better topic separation. Finally, the topic model with the highest consistency score was chosen. Our analysis revealed that the consistency scores for K = 34 reached the maximum C_V =0.6982, so we decided that 34 was the best number of topics [11,46]. As a result of the topic modeling implemented with the above-mentioned dataset and parameters, we created 34 topics. For each topic, the model displays the following data: (1) descriptive keywords: a list of the first N words with the highest frequency describing the topic, along with the probability of each word indicating the relative descriptive power of the word for the topic. We identify the top 20 words for each topic. (2) List of related articles: each related article is assigned a correlation value between 0 and 1. The higher the correlation of an article, the more “on-topic” the article is described. Following previous research, we assign the article with the highest correlation values to that topic. After the articles were assigned to the dominant topics, the labels for each topic were determined and assigned by three field experts, taking into account the descriptive keywords of the topics. Finally, taking into account the context and background of the 34 topics identified as a result of the analysis, these topics were categorized and a systematic taxonomic map was created, modeling the research and practice perspectives of the DT field.

4. Results

4.1. Descriptive Analysis (RQ1)

In order to answer the RQ1 question, this section provides bibliometric characteristics of DT research.

Table 1. Annual distribution of DT articles.

Year	N	%
2022	1865	34.86
2021	1499	28.02
2020	933	17.44
2019	522	9.76
2018	244	4.56
2017	121	2.26
2016	77	1.44
2015	41	0.77
2014	26	0.49
2013	22	0.41
Total	5350	100.00

shows the numbers and percentages of articles by year. The low number of studies conducted in the first four years (2013–2016) is highlighted in Table 1. Despite an increase in the number of studies conducted during the process, the majority of the studies (80.32%) were conducted during and after the pandemic period (2020–2022).

Table 2. The most productive publication sources.

Source Title	N	%
Sustainability Switzerland	288	5.38
Technological Forecasting and Social Change	68	1.27
IEEE Access	61	1.14
Journal of Business Research	56	1.05
Applied Sciences Switzerland	52	0.97
Frontiers in Psychology	38	0.71
IEEE Transactions on Engineering Management	30	0.56
International Journal of Environmental Research and Public Health	27	0.50
Sensors	27	0.50
Journal of Medical Internet Research	26	0.49
International Journal of Innovation Management	25	0.47
Technology in Society	25	0.47
Energies	24	0.45
Journal of Open Innovation Technology Market and Complexity	24	0.45
Journal of Manufacturing Technology Management	23	0.43

lists the most productive journals for DT. The majority of the DT articles were published in the journal Sustainability Switzerland, as shown in Table 2. This journal is followed by Technological Forecasting and Social Change and IEEE Access.

Table 3. Distribution of DT articles by subject areas.

Subject Area	N	%
Business, Management, and Accounting	1876	35.07
Social Sciences	1692	31.63
Computer Science	1666	31.14
Engineering	1467	27.42
Economics, Econometrics, and Finance	589	11.01
Environmental Science	540	10.09
Decision Sciences	474	8.86
Energy	435	8.13
Medicine	362	6.77
Mathematics	230	4.30
Materials Science	207	3.87
Arts and Humanities	198	3.70
Psychology	186	3.48
Physics and Astronomy	167	3.12
Chemical Engineering	116	2.17
Biochemistry, Genetics, and Molecular Biology	106	1.98
Chemistry	80	1.50
Earth and Planetary Sciences	80	1.50
Agricultural and Biological Sciences	79	1.48
Health Professions	74	1.38
Multidisciplinary	40	0.75
Pharmacology, Toxicology, and Pharmaceutics	32	0.60
Nursing	24	0.45
Immunology and Microbiology	19	0.36
Neuroscience	16	0.30
Dentistry	5	0.09
Veterinary	4	0.07

shows which subject areas are covered by the DT studies. Because the articles cover a wide range of topics, the sum of the distribution percentages exceeds 100% in Table 3. Further, this table demonstrates the multidisciplinary nature of DT. According to this table, the fields of “Business, Management, and Accounting”, “Social Science”, and “Computer Science” stand out in DT studies.

The keywords most frequently used in DT research are shown in

Table 4. Top 30 most used keywords.

Keywords	N	%
Digital Transformation	2514	46.99
Digitalization	527	9.85
Industry 4.0	431	8.06
Digitization	333	6.22
Artificial Intelligence	302	5.64
Innovation	280	5.23
COVID-19	253	4.73
Digital Technologies	246	4.60
Metadata	212	3.96
Sustainability	203	3.79
Manufacturing	188	3.51
Internet of Things	180	3.36
Big Data	179	3.35
Decision Making	166	3.10
Digital Economy	157	2.93
Sustainable Development	152	2.84
Digital Technology	143	2.67
Machine Learning	130	2.43
Pandemic	106	1.98
Information Technology	105	1.96
Blockchain	104	1.94
E-learning	104	1.94
Higher Education	103	1.93
Automation	94	1.76
Competition	91	1.70
Information Management	89	1.66
Education	83	1.55
Health Care	83	1.55
Social Media	81	1.51
Technological Development	79	1.48

. “Digital Transformation” is the most frequently used keyword, according to Table 4. “Digitalization” is the second most frequently used keyword, and “Industry 4.0” is the third.

Table 2 lists the most productive journals for DT. The majority of the DT articles were published in the journal Sustainability Switzerland, as shown in Table 2. This journal is followed by Technological Forecasting and Social Change and IEEE Access.

Table 3 shows which subject areas are covered by the DT studies. Because the articles cover a wide range of topics, the sum of the distribution percentages exceeds 100% in Table 3. In addition, this table demonstrates the multidisciplinary nature of DT. According to this table, the fields of “Business, Management, and Accounting”, “Social Science”, and “Computer Science” stand out in DT studies.

The keywords most frequently used in DT research are shown in Table 4. “Digital Transformation” is the most frequently used keyword, according to Table 4. “Digitalization” is the second most frequently used keyword, and “Industry 4.0” is the third.

4.2. Topic Modeling Analysis (RQ2)

The topics identified by LDA and their citation rates are presented in this section. The LDA topic model was applied to the DT experimental corpus, and 34 topics were discovered, as shown in

Table 5. The 34 topics discovered by LDA.

No	Topic Name	Rate (%)	No	Topic Name	Rate (%)
1	Management	6.13	18	Internet of Things	2.47
2	DT in Economy	5.88	19	COVID-19	2.45
3	SDLC	5.63	20	Service Provider	2.40
4	DT in Education	5.29	21	Smart Grid	2.28
5	Social Media	4.19	22	DT in Construction	2.17
6	DT in Health	4.15	23	Digital Literacy	2.09
7	Performance Analysis	3.79	24	HRM	2.09
8	Economic Policy	3.77	25	Supply Chain	2.02
9	Business Model	3.75	26	Artificial Intelligence	1.96
10	DT of SME	3.41	27	Cybersecurity	1.96
11	Digital Technologies	3.18	28	Cloud Services	1.94
12	E-Government	3.16	29	Virtual Reality	1.83
13	Industry 4.0	3.12	30	DT in Finance	1.81
14	Innovation Capability	3.01	31	DT in Tourism	1.45
15	Sustainable Energy	2.91	32	CIO-CDO Roles in DT	1.45
16	E-Commerce	2.80	33	Smart City	1.43
17	Big Data	2.55	34	Cultural Entrepreneurship	1.43

. Topic ratios expressed as percentages indicate the frequency with which they occur in articles. Appendix A,

Table A1. Keywords and percentages of the topics.

Topic Name	Keywords	%
Management	organization management organizational process change knowledge practice project strategy transformation strategic manager leadership organization role theory support interview leader need	6.13
DT in Economy	digitalization digital development economy process technology economic information problem system management analysis state modern level enterprise main activity consider determine	5.88
SDLC	system process model design quality management framework develop support information development evaluation software tool implementation improve project assessment requirement integrate	5.63
DT in Education	education learn student university higher educational teacher online school teach institution learning skill course teaching training technology e-learning support environment	5.29
Social Media	social media communication society theory news journalism content community platform discourse political practice online interaction network change transformation actor argue	4.19
DT in Health	health care healthcare patient medical clinical hospital medicine record disease implementation system service electronic include information outcome improve conclusion quality	4.15
Performance Analysis	performance effect relationship impact factor model influence data analysis leadership positive test role structural variable equation affect sample orientation investigate	3.79
Economic Policy	country economy economic policy european ict sector growth global regional region development level market eu trade analysis international index impact	3.77
Business Model	business model company strategy management industry customer market strategic process change competitive advantage innovation transformation sale experience corporate need product	3.75
DT of SME	sme adoption maturity factor enterprise readiness model level small technology digitalization medium-sized develop barrier company success adopt support framework assess	3.41
Digital Technologies	digital technology transformation change information development role emerge society need age develop infrastructure create transform challenge digitization opportunity require innovation	3.18
E-Government	public government governance policy service sector administration legal e-government regulation law citizen tax audit state private need electronic national authority	3.16
Industry 4.0	industry industrial manufacturing manufacture production technology revolution company process product lean factory smart implementation system sector development environment manufacturer challenge	3.12
Innovation Capability	innovation capability firm dynamic digitalization framework business technological knowledge process digital theory role resource organizational collaboration develop open disruptive managerial	3.01
Sustainable Energy	sustainable enterprise sustainability development environmental energy impact efficiency green effect innovation improve mechanism promote economy economic model consumption achieve analysis	2.91
E-Commerce	customer marketing online retail mobile consumer intention service experience user acceptance perceive app satisfaction quality brand market behavior retailer purchase	2.80
Big Data	data big analytical analysis information collect mining source data-driven collection database predictive available decision analyze challenge generate open time large	2.55
Internet of Things	internet iot technology thing blockchain smart agriculture application system agricultural sector data food challenge solution adoption farm sensor distribute agri -food	2.47
COVID-19	pandemic crisis change impact resilience time response global disruption world accelerate remote event emergency work increase rapid coronavirus situation challenge	2.45
Service Provider	service value platform ecosystem creation model servitization network create customer provider proposition share design delivery framework co-creation capture actor role	2.40
Smart Grid	network energy power system device communication application grid time control wireless sensor smart mobile frequency achieve demand filter electricity electric	2.28
DT in Construction	construction twin bim engineering model management information building industry maintenance project application build control operation site water infrastructure technology modeling	2.17
Digital Literacy	survey literacy perception professional interview qualitative data technology attitude group conduct analysis work level participant survey people gender social	2.09
HRM	work employee human skill resource job management professional workplace change worker talent competency workforce accounting role demand labor capital individual	2.09
Supply Chain	chain supply management logistics circular procurement economy supplier industry sc technology integration food material information value firm dsc performance digitalization	2.02
Artificial Intelligence	intelligence artificial machine automation learn technology process robotic knowledge human intelligent learning application system ml ethical cognitive robot deep decision	1.96
Cybersecurity	risk security information cyber cybersecurity threat privacy gas oil management protection data asset system dx attack personal disaster organization control	1.96
Cloud Services	cloud service architecture compute library enterprise information computing resource user algorithm framework application erp center global ea board solution web	1.94
Virtual Reality	virtual reality design lab laboratory technology ar device augment game fashion vr tool interactive application experience simulation augmented interaction environment	1.83
DT in Finance	financial bank banking fintech insurance service payment market finance risk institution transaction vietnam system technology sector commercial chatbot investment industry	1.81
DT in Tourism	tourism port sector maritime travel destination industry tourist hotel spanish hospitality ship spain startup brazilian international start-up sustainability container transport	1.45
CIO-CDO Roles in DT	control information price officer report global chief market rm cio stock cost investment cdo switch warehouse condition firm stability return	1.45
Smart City	smart city urban rural pharmaceutical area development plan service spatial space village drug pss pattern human activity pharmacy citizen environment	1.43
Cultural Entrepreneurship	entrepreneurship cultural entrepreneurial museum heritage art culture forensic entrepreneur story archive collection online work source experience evidence institution creation film	1.43

lists the descriptive keywords for each topic. The keywords for each topic are listed in descending order of frequency in Table A1. The top three topics with the highest percentages, as shown in Table 5, are “Management”, “DT in Economy”, and “SDCL” (software development life cycle). “CIO-CDO Roles in DT”, “Smart City”, and “Cultural Entrepreneurship” received the lowest percentages.

Table 6. Citation rates of the topics.

No	Topic Name	CR (%)	No	Topic Name	CR (%)
1	Management	9.67	18	Economic Policy	2.37
2	Business Model	8.43	19	Service Provider	2.32
3	Industry 4.0	6.53	20	DT in Construction	1.84
4	Digital Technologies	6.31	21	Big Data	1.83
5	Innovation Capability	5.41	22	CIO-CDO Roles in DT	1.67
6	COVID-19	4.16	23	Smart Grid	1.63
7	DT in Health	4.08	24	Artificial Intelligence	1.58
8	Social Media	3.99	25	Sustainable Energy	1.47
9	SDLC	3.60	26	HRM	1.35
10	DT in Education	3.52	27	Smart City	1.24
11	Performance Analysis	3.45	28	Cybersecurity	1.17
12	DT of SME	3.44	29	DT in Finance	1.08
13	Internet of Things	2.92	30	Cloud Services	1.05
14	E-Government	2.61	31	Virtual Reality	0.96
15	DT in Economy	2.58	32	Cultural Entrepreneurship	0.93
16	Supply Chain	2.58	33	DT in Tourism	0.91
17	E-Commerce	2.56	34	Digital Literacy	0.74

also shows the citation rates (CR) of the discovered topics. The “Management” topic with the highest percentage is also the most cited topic, according to Table 6. This is followed by the “Business Model” and “Industry 4.0” topics. Although the percentages of both topics are low (Table 5), the citation rates are high. Similarly, “Digital Technologies”, “Innovation Capability”, and “COVID-19” topics have high citation rates despite having a low percentage. “Cultural Entrepreneurship” and “DT in Tourism” have the lowest citation rates as the topics with the lowest percentages. “Digital Literacy”, on the other hand, stands out as the topic with the lowest citation rate.

4.3. Taxonomy of Digital Transformation Topics (RQ3)

The topics were classified, and a taxonomy map was created in this section, which included four major categories: implementation, process, technology, and people (see

Table 7. Taxonomy of the DT topics.

No	Sub-Category	Topic	Rate %	Total %
1	Implementation	DT in Economy	5.88	45.76
		DT in Education	5.29
		DT in Health	4.15
		DT of SME	3.41
		E-Government	3.16
		Industry 4.0	3.12
		Sustainable Energy	2.91
		E-Commerce	2.80
		COVID-19	2.45
		Smart Grid	2.28
		DT in Construction	2.17
		Supply Chain	2.02
		DT in Finance	1.81
		DT in Tourism	1.45
		Smart City	1.43
		Cultural Entrepreneurship	1.43
2	Process	Management	6.13	26.08
		SDLC	5.63
		Performance Analysis	3.79
		Economic Policy	3.77
		Business Model	3.75
		Innovation Capability	3.01
3	Technology	Social Media	4.19	22.48
		Digital Technologies	3.18
		Big Data	2.55
		Internet of Things	2.47
		Service Provider	2.40
		Artificial Intelligence	1.96
		Cybersecurity	1.96
		Cloud Computing	1.94
		Virtual Reality	1.83
4	People	Digital Literacy	2.09	5.63
		HRM	2.09
		CIO-CDO Roles in DT	1.45

). As shown in Table 7, the first category is implementation (45.76%). The implementation category includes specific areas where digital transformation is used.

This category contains 16 topics, including “DT in Economy”, “DT in Education”, “DT in Health”, “DT of SME”, “E-Government”, “Industry 4.0”, “Sustainable Energy”, “E-Commerce”, “COVID-19”, “Smart Grid”, “DT in Construction”, “Supply Chain”, “DT in Finance”, “DT in Tourism”, “Smart City”, and “Cultural Entrepreneurship”. The second category, process (26.08%), includes elements such as developing, understanding, improving, implementing, and managing organizational business processes for a successful DT. The process category contains six topics: “Management”, “SDLC”, “Performance Analysis”, “Economic Policy”, “Business Model”, and “Innovation Capability”. The third category is technology (22.48%). This category is about DT technologies and tools. “Social Media”, “Digital Technologies”, “Big Data”, “Internet of Things”, “Service Provider”, “Artificial Intelligence”, “Cybersecurity”, “Cloud Computing”, and “Virtual Reality” are all topics in the technology category. People (5.63%) are the fourth category, and they play an important role in DT. Topics in the people category cover the roles of human expertise in organizational DT. “Digital Literacy”, “HRM” (Human Resources Management), and “CIO-CDO Roles in DT” are the topics.

4.4. Time-Based Trend Analysis of the Topics (RQ4)

To reveal the changes in trends and evolutions of 34 topics reflecting the DT research landscape over time, annual changes in percentages of topics were identified, and trend values for each topic were calculated. Appendix A,

Table A2. Annual percentages and trends of the topics.

Topic Name	2013	2014	2015	2016	2017	2018	2019	2020	2021	2022	Trend
Sustainable Energy	0.00	0.00	0.00	0.00	0.00	0.00	0.85	0.48	2.39	5.90	1.18
DT in Health	8.33	6.25	6.67	12.07	0.93	2.33	5.31	4.44	3.43	4.33	0.68
E-Government	0.00	6.25	3.33	1.72	0.00	3.26	3.61	4.08	3.28	2.71	0.54
DT in Education	8.33	0.00	6.67	5.17	2.78	3.26	5.73	4.80	5.82	5.42	0.53
Supply Chain	0.00	6.25	3.33	0.00	0.00	0.93	0.64	2.04	2.31	2.47	0.49
Innovation Capability	0.00	0.00	3.33	1.72	0.93	1.86	1.27	2.64	3.88	3.37	0.49
Performance Analysis	0.00	0.00	0.00	0.00	2.78	1.40	2.76	2.52	4.03	5.18	0.48
COVID-19	0.00	0.00	0.00	3.45	0.00	1.86	0.85	3.60	3.06	2.11	0.42
DT of SME	0.00	0.00	0.00	0.00	1.85	2.79	1.49	3.96	3.81	3.79	0.39
Digital Literacy	0.00	6.25	3.33	0.00	0.93	2.79	0.64	1.92	1.79	2.83	0.38
Internet of Things	0.00	0.00	0.00	3.45	0.93	2.33	1.49	2.88	2.54	2.65	0.34
DT in Construction	8.33	0.00	0.00	0.00	0.93	1.40	2.34	1.92	2.24	2.47	0.31
E-Commerce	8.33	0.00	10.00	1.72	1.85	2.33	2.97	2.40	2.39	3.31	0.29
Management	0.00	6.25	6.67	6.90	4.63	6.51	5.31	6.12	6.57	6.08	0.29
Artificial Intelligence	0.00	6.25	0.00	0.00	0.93	1.86	2.12	1.20	2.24	2.23	0.26
Economic Policy	0.00	0.00	0.00	1.72	2.78	0.93	4.25	4.08	3.88	4.03	0.25
HRM	0.00	0.00	0.00	1.72	0.93	1.40	1.49	2.40	2.39	2.11	0.24
Service Provider	8.33	0.00	0.00	0.00	1.85	2.33	2.34	2.52	1.94	2.89	0.21
DT in Tourism	8.33	0.00	0.00	0.00	0.93	1.86	0.42	1.20	1.64	1.75	0.16
DT in Finance	0.00	6.25	0.00	0.00	0.93	2.33	1.91	2.64	1.87	1.38	0.09
Cybersecurity	0.00	0.00	0.00	3.45	1.85	1.86	1.49	2.52	1.72	2.05	0.04
SDLC	0.00	0.00	6.67	1.72	6.48	6.51	4.88	5.40	5.52	6.08	−0.08
Smart Grid	16.67	6.25	0.00	5.17	2.78	2.79	1.49	1.92	2.46	2.23	−0.11
CIO-CDO Roles in DT	0.00	6.25	0.00	6.90	1.85	2.79	1.91	1.20	1.19	1.26	−0.12
Virtual Reality	0.00	0.00	0.00	0.00	2.78	1.86	3.18	1.08	1.49	2.17	−0.12
DT in Economy	0.00	0.00	0.00	0.00	4.63	5.12	9.13	8.16	6.42	3.97	−0.13
Cultural Entrepreneurship	8.33	12.50	6.67	3.45	1.85	4.19	1.91	1.20	1.19	0.90	−0.19
Smart City	0.00	0.00	0.00	0.00	2.78	3.72	2.34	1.20	1.79	0.72	−0.41
Cloud Services	0.00	6.25	3.33	10.34	4.63	3.26	1.91	1.68	1.94	1.38	−0.65
Business Model	0.00	0.00	0.00	10.34	5.56	5.58	6.79	4.08	3.73	2.29	−0.65
Industry 4.0	0.00	0.00	0.00	0.00	6.48	4.65	2.76	4.44	2.54	2.83	−0.73
Big Data	0.00	0.00	3.33	5.17	7.41	4.19	3.61	2.40	1.87	2.29	−1.02
Digital Technologies	0.00	6.25	13.33	6.90	10.19	5.58	5.52	3.00	2.91	1.75	−1.69
Social Media	25.00	18.75	23.33	6.90	13.89	4.19	5.31	3.84	3.73	3.07	−2.16

“In this table, the green colors indicate percentages above the topic’s average in a row, the red colors indicate percentages below the topic’s average, and their colors approach white as these annual values approach the average value”.

displays the percentages, trend values, and trend directions of the topics by year. The topics in this table are listed in descending order (from green to red) based on the overall trend values in the last column (see Appendix A, Table A2). Annual percentages for the last five years were used to calculate the overall trend shown in the last column, because in the first five years, the percentage weight of all the topics has not been formed yet. Appendix A, Table A2 is presented in the form of a colorized heatmap table. Each row of this table demonstrates the annual percentage changes for a specific topic as well as descriptive information about how that topic has evolved over time. In this table, the green colors indicate percentages above the topic’s average in a row, the red colors indicate percentages below the topic’s average, and their colors approach white as these annual values approach the average value (see Appendix A, Table A2). Figure 2 also depicts the top ten topics with the highest increasing and decreasing trends. According to Figure 2, “Sustainable Energy” appears to be the topic with the highest increasing trend. This topic is followed by “DT in Health”, “E-Government”, “DT in Education”, and “Supply Chain”, respectively. “Big Data”, “Digital Technologies”, and “Social Media” have the most decreasing trends. On the other hand, despite the low percentage of topics (see Table 4) and citation rate, the topic “Sustainable Energy” shows an increasing trend (see Table 5). “Social Media” is another fascinating topic in this context. The topic “Social Media” has the most decreasing trend, despite having a relatively high percentage of topics and citations.

5. Discussion

5.1. The Interdisciplinary Background of Digital Transformation

DT, the focal point of today’s technological life cycle, has common application areas for all disciplines. Technological advancements, digital infrastructure, and globalizing business models enable large-scale DT projects based on interdisciplinary collaboration by bringing together multi-disciplinary approaches and ideas from disparate disciplines. DT is a process that affects all disciplines and has various application areas based on the collaboration of various disciplines in each sector [1,6]. Business (Industry 4.0, Business Intelligence, Smart Innovation), Health (E-Health, Smart Drug Management), Energy (Smart Grids, Cloud Energy, Sustainable Energy), Technology (Virtual Reality, Big Data, Machine Learning, Deep Learning, IoT), and so on are examples of common applications [3,6].

In line with previous studies, our findings confirmed that DT has an interdisciplinary structure. Furthermore, the contribution rates of other disciplines contributing to the field are shown in our findings (Table 3). According to our findings, the fields of “Business, Management, and Accounting”, “Social Sciences”, “Computer Science”, and “Engineering” made the greatest contribution [6]. This multidisciplinary background of DT is clearly reflected in our findings [6,47,48]. For example, “Management”, “Business Model”, and “DT in Finance” topics are related to business, management, and accounting, “DT in Education” and “E-Government” topics are related to social sciences, “SDCL”, “Big Data”, “Artificial Intelligence”, and “Internet of Things” topics are related to computer science, and “Smart Grid” and “DT in Construction” topics are related to engineering. Our findings regarding the multidisciplinary nature of DT (economics, health, education, management, etc.) can be considered as an indication that interdisciplinary collaboration produces more innovative and effective solutions for DT processes [6].

5.2. The Four Pillars of Digital Transformation

For the 34 topics obtained as a result of this study, a systematic taxonomy with four basic categories was created. In a systematic taxonomy, DT is divided into four major categories: application, process, technology, and people. Within these four categories, DT has an interacting life cycle and a working flow that combines various elements working together. The categories we presented are similar to those presented in previous studies. The dominance of process, technology, and people categories in DT is a phenomenon that researchers have previously highlighted [24]. In addition to these categories, the study created the implementation category, which evaluates DT in the context of industrial applications.

As previously stated, the spread of digital technologies, the pandemic, and other similar factors have accelerated global DT. As a result of this acceleration, organizations, industries, and government agencies were forced to transform their existing business models in order to keep up with the changing world ushered in by information technologies [49]. The transformation of a business model that occurred in one industry triggered the participation of other industries in this transformation. This is supported by our findings in the implementation category. The implementation category accounts for nearly half of the total category weights (45.76%). DT applications are used in a variety of fields, including the economy, manufacturing, energy, and cultural entrepreneurship. As a result, the spread of field applications to a broad spectrum is considered a natural result. Furthermore, the most striking topic of the implementation category (Table 7) is “DT in Economy”, which is ranked second in the order of topic percentages (Table 6). This situation indicates that there is scientific interest in comprehending the theoretical and practical aspects of the economy’s DT in general, as well as its implications for economic units [49]. It can be seen that the other prominent implementation category topics are “DT in Education” and “DT in Health”, which reflect sectors that can be considered a start in DT (see Appendix A, Table A2). The fact that researchers are turning to the applications of these industries can be viewed as a natural result of the widespread impact of education and health on society [50].

Processes are critical paradigms for improving workflow efficiency [26]. By definition, DT entails either improving existing processes or developing new, more efficient processes. As a result, effective process management is closely related to successful DT [51]. Because of this relationship, the “Management” topic is the most important topic in the process category. Furthermore, the fact that the topic “Management” ranks first in terms of frequency of occurrence (Table 5) and citation rates (Table 6) supports this finding. Our findings, on the other hand, show that the “SDLC” topic ranks first in both the process category and the frequency of occurrence. This finding indicates that businesses that previously did not focus on software but produced basic products and services are now adopting software-oriented services and applications more frequently [52]. Furthermore, the fact that software is one of the most important driving forces for DT explains why the “SDLC” topic is at the top of the list [52,53].

Technology is a tool that can help organizations in their DT process [27]. Social media is an effective subset of this tool. Social media has capabilities that are important for DT, such as improving stakeholder communication, education, and information sharing. Furthermore, it is less expensive than other DT technologies [54]. According to studies, the use of social media accelerates the DT of institutions ranging from small businesses to government agencies [55,56]. The fact that “Social Media” is the most important topic in the technology category confirms this situation. Another intriguing finding in the technology category is that the topic “Virtual Reality” is ranked last. Virtual reality is one of the rapidly evolving, low-cost, and widely available technologies [57]. However, our findings indicate that virtual reality technology is only just getting started in the context of DT.

Employees play an important role in achieving the goal of DT in an organization [25]. As a result, understanding employee roles and skills is critical for DT. As a result, there is a direct relationship between employee digital literacy and the success of DT. Digital literacy is increasingly being used in practice and theory to address employees’ knowledge, skills, and ability to interact with digital technologies in this context [58,59]. Our discovery that the most important human category topic is “Digital Literacy” emphasizes this point. Human–machine collaboration, on the other hand, must be effectively ensured for a successful DT. In this regard, close collaboration between the IT and HR departments will significantly aid the organization’s DT [60]. As can be seen from this, the active role of human resources in DT necessitates placing the “HRM” topic at the top of the human category. The broad spectrum of DT’s applications indicates its widespread adoption in various industry and public domains. For the DT process to be carried out successfully in organizations, it is necessary to consider DT holistically, not only from a technological standpoint, but also by evaluating the process and human mechanisms.

5.3. Existing Trends in Digital Transformation from Past to Present

The COVID-19 pandemic has been a driving force for digital transformation, particularly in critical areas of society. In other words, the COVID-19 outbreak has accelerated the trend toward topics directly related to social life, such as “Sustainable Energy”, “DT in Health”, “E-Government”, “DT in Education”, and “Supply Chain” [61,62]. Energy, in particular, plays an important role in the global dimension and is central to many sustainable development goals that are closely related to societies [63]. As a result of the pandemic, the DT of energy has become critical for the development of sustainable energy systems [64]. One of the most effective ways to address current issues such as energy crises and climate change is to ensure the long-term viability of clean, renewable energy. All of these arguments have distinguished “Sustainable Energy” for researchers. The fact that the topic “Sustainable Energy” is the fastest growing topic can also explain the critical developments in the energy field.

Another significant finding of the study is slowdown in the momentum of “Social Media”, “Digital Technologies”, and “Big Data” topics over the last five years (see Figure 2). The increase in the diversity of DT’s research contexts and application areas in recent years has caused the core topics studied in the first years to be divided into sub-contexts, and as a result, a slowdown has been observed in such core topics of DT. Although there was a lot of emphasis on these topics in the early stages of DT, the COVID-19 outbreak resulted in a rapid decline in interest in these topics (see Appendix A, Table A2). Furthermore, “Industry 4.0” was identified as one of the low acceleration topics. Unlike previous industrial revolutions, Industry 4.0 has a technology-driven connection to the changes that have occurred [65]. As a result, it is believed that the low acceleration in the “Industry 4.0” topic is due to trends in post-COVID-19 technology topics [66].

6. Conclusions

This research aims to reveal the research and application topics of DT, as well as their temporal trends and hierarchical taxonomy. Semantic content analysis was performed on 5350 peer-reviewed journal articles specific to the field for this purpose, using the LDA topic modeling algorithm, which is commonly used in text mining. The research themes of the field, their taxonomy, and their evolution over time were determined using this analysis. According to the topic percentages and citation rates of topics in our study, transformation management holds an important place in DT research. As another finding, four subcategories of DT were identified: implementation, process, technology, and people. More specifically, our taxonomy findings revealed that nearly half of the studies covered DT implementations in various fields. Furthermore, contrary to expectations, our research found that there is little interest in virtual reality technology, which is one of the most popular technologies. Furthermore, when the number of topics, the frequency of occurrence of the topics, and citation rates are considered, it is possible to conclude that the human component of DT has not yet received sufficient attention. Our findings also show that internal stakeholders are generally focused on the human component, while research trends for external stakeholders are low. Furthermore, our research shows that in the last five years, researchers have become more interested in industrial applications of DT. In this respect, “Sustainable Energy”, “DT in Health”, “E-Government”, “DT in Education” and “Supply Chain” were identified as topics with the most increasing trend. Another important finding from our research is that DT research, particularly in energy, is still in its early stages and evolving rapidly.

This study, like all others, has limitations. In this study, only peer-reviewed journal articles were used. The study can be repeated in future studies by including different types of studies, such as books and conference papers from various databases. Furthermore, different topic modeling algorithms can be used to perform analyses, and the results can be compared. More experimental research is needed for comparison and validation of our results in the field of DT. It is expected that our methodology and findings will be useful in future research. Furthermore, this study is expected to serve as a guide for researchers to investigate potential opportunities in the field. One of the potential benefits of this study is that it will aid strategic decision-making processes by increasing practitioners’ awareness of DT.