Abstract
Despite growing interest in enterprise architecture (EA) around the world in recent years, a lack of common understanding is frequently described by EA researchers/practitioners. We conducted a systematic mapping study and it revealed that the extent to which the authors/researchers are focused on EA, the sectors in which they are working, the academic disciplines in which they have studied, the countries where their affiliated organizations are located, the subject areas of the journals/publishers of their publications and the way they have approached EA and its practitioners are some major elements that might influence the existing uniformity in EA. In addition, this study demonstrates how important it is to pay attention to the definition of ‘enterprise architecture’ itself. The contribution of this study is the organization of the EA literature according to three major questions concerning ‘who’ have been published in the literature, ‘where’ they have been located and ‘what’ their publications are about. This helps to better identify sources of variety which could be on the basis of the lack of common understanding in EA and provides practitioners and stakeholders a better understanding of this challenge. This also provides relevant directions for future studies.
Keywords
Introduction
Contemporary organizations regularly encounter challenges meeting their Information Technology (IT) needs, be it a simple tool with which to save and organize data, an indispensable strategic and competitive weapon or unique routine administrative tasks, such as decision-making that needs fulfilling. 1 According to some researchers, enterprise architecture (EA) is the discipline and practice that emerged in order to help organizations meet these challenges 2 in order to survive in an increasingly dynamic environment full of interruptions and change.
EA has generated growing interest in recent years, as shown by the numerous scientific articles published by EA researchers and practitioners, EA conferences organized around the world and new frameworks built to improve EA practice. But researchers and practitioners have described a serious lack of ‘uniformity’ in EA, as presented in Table 1, in spite of this significant progress.
Expression of the existence of various perspectives on EA.
EA: enterprise architecture.
The lack of ‘uniformity’ in EA is also presented in the study by Lapalme 4 as the ‘existence of many ways to approach EA’. This study intends to identify the elements in the literature that can play a role in this challenge that EA is facing. To achieve this objective, this article systematically selected and reviewed the EA literature by following a few research questions.
The research problem and literature review are presented in ‘Background’ section. ‘Research design’ section presents the research questions and the methods applied to examine these questions. The results and their discussions are presented in ‘Contextualization of the findings’, ‘Quantitative findings’ and ‘Qualitative findings’ sections. And ‘Discussions’ and ‘Conclusion and future work’ sections present some discussions concerning the findings and useful directions for future work.
Background
EA literature lacks uniformity of definition as well as a description of the term ‘enterprise architecture’ itself. 9 The definitions of EA vary in terms of ‘scope and purpose’. 4 This situation can create misunderstanding and conflict regarding the role and responsibility of professionals practicing EA, especially when EA team members are not thoroughly conscious of the extent of the lack of common understanding in EA. It can also be hard to collaborate with stakeholders and other participants in such situations. Similarly, this makes it hard to provide standard and universal training to future EA practitioners. EA researchers can face difficulty effectively sharing their findings and generally being understood.
Such problems represent a few complications experienced by researchers and practitioners. This is why some have reported that EA is an ‘immature practice’ 7,10 suffering from a ‘lack of common terminology’ 3 and ‘shared meaning’, 4 and EA literature is facing a challenge of ‘fragmented discourse’. 5 As a matter of fact, this issue concerning the terminological differences in EA has been mentioned in the publications of many researchers, even if it is not the main focus of their work. Others have investigated this issue more thoroughly and came to more accurate conclusions. To achieve this, they reviewed and analysed the EA literature and surveyed researchers and practitioners. 10 –12 In a similar way of identifying various terminology and perspectives in EA, some previous works affirm the existence of three schools of thought in EA. 4 This work has compared EA to an Indian parable which describes how six blind men who touched an elephant for the first time perceived it very differently – depending on the part of the body they happened to touch. This comparison contributes to awareness-raising conversations concerning the various ways of approaching EA, and therefore allows for the opportunity for EA to become more mature as a field through the establishment of a common structure.
Even though a large number of studies have mentioned this lack of common understanding in EA, only a few of them have realized a deeper investigation of the problem and employed a rigorous methodology to conduct their analysis. 2 –7 A few formal systematic mapping studies (SMSs) and systematic literature reviews (SLRs) also exist on EA. Moreover, the rest of this section presents some existing literature reviews on EA.
In fact, 6 conducted a state-of-the-art review from 1987 to 2010 in order to investigate the collaboration of scholars in EA management via co-authorships and its impact on the diffusion of their contributions. They also investigated the main EA research streams, their interlink and the major works to be assigned to these streams. And finally, they investigated the focus concerning specific dimensions of EA research content (layer, methodology, task and life cycle).
On the other hand, 11 used bibliographic analysis standard tools to study EA within the public administration from 1999 to 2014 and investigated the publishers and their subject areas, the authors of the publication, the correlations among the keywords, the definitions of EA in public administration, government EA programmes around the world and so on.
On the other hand, 13 conducted a SLR on EA in the public sector from 2005 to 2014, which investigated the main topics of the EA publications, their themes, their geographical distribution, the research methods used and the number of citation.
On the other hand, 14 conducted a general SLR on EA from 2000 to 2015, which investigated the publishers of the papers and their topic, the authors and the country of their affiliated organizations.
However, none of the previous literature reviews focused on the whole discipline of EA and its lack of common understanding. Consequently, there is a need for literature reviews which further our understanding of this lack. This investigation is intended as an input that might contribute to fill this gap, 15 by conducting a SMS 16 using articles published from 1990 to mid-2018 in major engineering, computer science and management journals.
Research design
Introduction to SMS
A frequent approach used to review and analyse literature in order to ‘realize a complete overview of a research area’ is SMS. SMS can contribute by finding ‘whether research evidence exists or not’. 17 When research evidence exists on a topic, SMS can also provide indicators of its reliability. The process involves performing a systematic classification of literature and its interpretation. The categories generated with this systematic classification are based on pertinent data that include, for example, information concerning the authors and publications – such as authors’ names, authors’ affiliations, authors’ country, publication sources, publication type and publication chronology – and information concerning the research design and research techniques employed to conduct studies and generate the findings. 16 The outcome of an SMS provides mainly a complete list of publications on the topic area investigated, presented in the form of classification where distinct categories are identifiable. 18
SLR is another methodology that has frequently been used to review and analyse the literature of a field in order to provide relevant directions for future investigations. But SMS and SLR do not analyse the literature in the same way. SMS can help to structure a research area, while SLR can help to gather and synthesize evidence. 18 SMS frequently answer a large amount of research questions. For example, this study includes nine research questions. To achieve this, SMS ‘collects data from the literature with sufficient detail and summarizes them with respect to many defined categories’, whereas SLR examines to what extent the research findings of each publication are consistent or inconsistent in order to ‘answer only a few specific research questions’. 16 However, the results of a previous SMS can be extremely useful in order to determine appropriate areas for conducting a relevant SLR. 16
Motivation to conduct an SMS
A systematic examination like SMS can greatly help identify elements from which the many ways to approach EA have originated or simply the existing different ways to approach EA. In fact, the use of SMS as a rigorous methodology to conduct this study will enhance its data selection, its data extraction and its analysis process. The use of SMS will also increase the reliability of this study’s findings.
Definition of research questions
According to the guidelines of Kitchenham et al. 16 and Petersen et al., 18 the first task of SMS is to ‘define the research questions’. The research questions indicate the scope of the study and specify what aspect it takes or does not take into account. 16
This SMS investigates the following nine research questions, classified in three categories as enumerated in Table 2. The intent is to identify the different ways to approach EA, to investigate which characteristics contribute to the existence of these different ways to approach EA and to understand how the EA community has become aware concerning this situation.
Research questions and rationales.
EA: enterprise architecture.
Conducting the search for primary studies
The second task is to create a data search strategy that can help to ‘identify and locate reliable data sources which can be used to extract the information to be analyzed’. 16,18
Because this study intends to provide a broad view of the discipline of EA, all the publications corresponding to EA should be significant to be analysed. With the objective to keep this research to a manageable size, only publications which explicitly mention EA or EA practitioners in their title were taken into account. The following search strings were appropriate to search publications:
‘enterprise architecture’ OR ‘enterprise architectures’ OR ‘enterprise architect’ OR ‘enterprise architects’ – in the Title.
Search was operated in the following electronic libraries: Compendex, Inspec, Scopus, IEEE, AIS and Google Scholar. These electronic libraries were considered because according to some previous searches, they are the libraries which have returned most of the major scientific publications with the article type selected and the search keywords used. They are also the libraries which are considered among the most relevant ones. 19
Table 3 presents the number of articles returned by each of the electronic libraries consulted. Google Scholar was often consulted for additional search and to download the full text of the articles.
Number of articles returned by the electronic libraries.
Screening articles based on inclusion/exclusion criteria
The third preoccupation of this SMS is to select only relevant data sources corresponding to the identified search strategy. 16,18 In fact, the results of each digital library were exported into BibTex (.bib) files. Software usable for SLR and SMS (StArt) were used in order to upload these data. After examining the titles, abstracts, introduction and conclusions of the identified articles, duplicate articles and articles without the aforementioned terms corresponding to EA research or practice were removed.
In addition, at the start, the articles selected were only those that were downloadable on the Internet with a licence from the authors’ affiliate libraries. However, other measures were also used when possible, in order to find copies of the articles, such as loans between university libraries and email contact with the authors of non-downloadable works.
With the objective to keep this research to a manageable size, ‘researchers can search only a targeted set of publications as data sources, and then restrict themselves to only one (1) publication type for example’. 16 This explains the choice to select only journal articles as data sources. Moreover, peer-reviewed articles were selected in order to stay focused on more professionally executed research.
Table 4 summarizes the complete criteria used in order to include the appropriate data sources before the search, and after reading the title, introduction and conclusion. The exclusion criteria correspond to the values that are different from those indicated in this table.
Inclusion criteria.
EA: enterprise architecture.
Because this study does not map a particular aspect on EA but aims to gather information concerning the lack of common understanding in EA, all the journal articles available which have met the condition indicated in Table 4 were included and no quality assessment stage was conducted.
Data extraction, analysis and classification
Another important preoccupation of SMS is to ‘create a classification scheme’. 17 Capturing ‘the state of the art’ in EA practice and research is the objective of our scheme. Because this study intends to have findings which really describe the situation of EA, it was not important to create a predefined classification scheme. A multifaceted classification scheme was consequently developed gradually, depending on the characteristics of the data collected.
In fact, the first author read entirely each article at least once, during which relevant data were collected. Most of the data collected were extracted as found, without any specific interpretation, in a MS Excel spreadsheet, in order to be able to format them automatically and to create the corresponding categories. The first author classified each article and applied a test–retest approach. The final classification was formally discussed many times with the second author.
After the publication of a first version of this study, many modifications were made to improve the study, including additional articles being were added. The data extraction process was realized by another person in accordance with a data extraction protocol that includes the categories found in the previous version, as presented in Table 5.
Summary of the data extraction protocol.
EA: enterprise architecture.
The last task of this SMS – without considering the report – is to ‘analyze and interpret the data extracted’ in the articles. 16,18 As can be seen in the column source of Table 5, the data extraction of certain information to collect required some analysis and attribution to a category. Furthermore, after collecting all the necessary information, various processes of data processing, such as validation, sorting, analysis and classification were applied in order to summarize the data collected. In the next sections, we present the different categories found, their occurrences and their similarity/dissimilarity compared to the other categories.
Validity evaluation
In terms of descriptive validity, the data extraction protocol used to extract and derive data from the articles allows the data extraction process to be objective because this process can be always re-examined.
In terms of theoretical validity, appropriate studies could not be identified during the search for primary studies. 18 To reduce the number of articles that have been missed, an additional search was conducted. In fact, few SMS exists on EA, yet it was not possible to compare the articles identified for this mapping study to others. But it was possible to compare these articles to those identified for an SLR which intended to summarize the existing work done in EA from 2005 to 2014, found with the strings ‘enterprise architecture’ either in the title, abstract or keywords. However, eight new articles – found in the study by Rasti et al., 13 in which an SLR intended to summarize the existing work done in EA from 2005 to 2014, – were added in the current study. Another strategy to reduce the bias was to conduct additional searches on Google Scholar and thus 10 articles were added. As a result, 257 articles were selected for examination. Table 6 presents the number of articles selected at each phase of the selection process.
Evolution of the number of articles selected.
Concerning the validity of the data extraction process, the articles were classified individually by two persons, but their classification was then reviewed and discussed. A test–retest approach was also applied on a sample because the first version of this study, which considered a limited number of articles, was previously published in a conference.
Table 7 presents a publication timeline of the 257 articles selected for examination.
Publication timeline of the articles selected for examination.
Contextualization of the findings
Distribution of the articles over the years
The 257 journal articles selected for examination were published over approximately 21 years, between 1990 and mid-2018. The year 2018 is absent in Figure 1 because the search was conducted in mid-2018, and so it was not possible to include all the EA journal articles published during this year. In effect, only nine articles were found for this year.
Journal article distribution by the publication year.
The distribution of the articles, as presented in Figure 1, demonstrates an absence of publications for the years 1991–1996, 1998 and 2002. Compared with the number of published articles in other disciplines as new as EA, the number of EA articles published over the years is few. However, this graph is still useful as it shows how EA has gained interest over the years.
Quantitative findings
What is the experience of EA researchers/authors?
Approximately 568 first and corresponding authors, including both researchers and practitioners, have contributed to the selected articles. Approximately 9% of these authors contributed between 3 and 7 articles, as presented in Table 8.
List of authors who have contributed to more than two articles.
Approximately 65% of the authors published only one of the articles. It would seem then that EA literature lacks publications from experienced researchers in the discipline.
What is the occupation of EA researchers/authors?
Figure 2 presents the occupation of the authors.
Occupation of the authors.
As seen in Figure 2, approximately 77% of the first and corresponding authors of the articles are ‘students or professors’ who come from schools, universities, faculties, institutes, research centres or laboratories.
Approximately 11% of these authors are ‘professional practitioners’ who come from private or public organizations, such as research agencies, government agencies and consulting firms.
Approximately 9% of these authors of the articles come from ‘both professional organizations and academia institutions’, and their research is based on partnerships between industry and academia.
Finally, because of a lack of information in the articles concerning the affiliation institution and no possibility of finding it on the Internet, the affiliation of 3% of these authors is considered as ‘unknown’.
A large majority of the articles selected derive from the academic world. This is to be expected because this study includes only scientific articles. But why have these articles presented many different ways to approach the discipline of EA as demonstrated in the following sections? It would seem then that EA lacks agreed references to follow in the academic world.
What is the focus of EA publishers/editors?
The selected articles were published across approximately 132 journals. Approximately 23 of these journals published 43% of the articles (as presented in Table 9) and represent the most significant publications, at 3–10 articles each. The editors and publishers of these journals include Taylor & Francis, Elsevier, Cutter Consortium, Springer Frontiers and IEEE, which are among the most well-known ones in the academic sector.
List of editors/publishers that contributed to more than two articles.
Journals may cover numerous subject areas. For example, one of the journals has 37 subject areas. The blank cells in Table 9 indicate cases in which it was not possible to find information concerning the subject area of the corresponding journal.
The classification of subject areas shows that a large majority of the journals correspond to subject areas related to Information Technology (i.e. computer science). It would seem that EA lacks editors/publishers dedicated specifically to EA publications.
What are the academic disciplines in which EA researchers/authors have studied?
The first authors of 87% of the selected articles came from academia. When considering the department, faculty, institute or laboratory where they conducted the research published in these articles, three main categories of study were identified.
Information Technology (IT): This category includes articles which indicate that the first authors are studying in Information and Communication Technology. It also includes authors who is studying in corresponding fields, like Informatics, Information Systems, Software, Computer Science or Computer Engineering; Social and human science (SS): This category includes articles which indicate that the first authors are studying in social fields like Administration, Management, Business, Economics, Communication Logistics or Marketing; and Specific area of engineering (SE): This category includes articles which indicate that the first authors are studying in a precise field of engineering different than Information Technology and its corresponding fields. Authors of this category are studying, for example, in Operation Research Mechanical, Electrical, System and Industrial. This category also includes the names of study that mixed several specific fields of engineering, like Industrial Information, Supply Chain Management, Mines-Telecom and Control Systems.
The absence of enough information concerning the study area of the first authors of some articles was a reason to consider the following other categories in addition to the previous ones. Non-identified areas of engineering (E): This category includes articles which indicate that the first authors are studying in a general name of study that might refer to several other specialized engineering fields. Some examples of the names of study put in this category are: the Faculty of Technology Engineering and Environment, the Faculty of Science and Engineering, the Department of Computer Science and Engineering and the Faculty of Technology and Engineering; Other (O): This category includes articles which indicate that the first authors are studying in a field different than IT, engineering and social sciences, as presented in the previous categories. This category includes two authors, one who is studying in a School of Medicine and the other in a Center of Forest Studies. Absent (ABS): This category includes articles which do not indicate enough interpretative information concerning the study area of the first authors. When this situation occurs, sometimes it is possible to find the study area of the authors on the Internet, in their other publications. But other times it is not possible to find this information.
Figure 3 presents the previous categories concerning the academic disciplines in which EA researchers/authors have studied, including a category N/A (non-applicable) for first authors who are not affiliated with an academic institution (professional) or when their sector of activities are absent.
Academic disciplines in which EA researchers/authors have studied. Information Technology (IT), social and human science (SS), specific area of engineering (SE), non-identified areas of engineering (E), other (O) and absent (ABS).
Where are the affiliated organizations of the first EA researchers/authors located?
The article distribution by country of publication shows that the affiliated institutions of the first authors are located in 46 countries. This also shows that a large majority of the articles come from institutions located in Europe, which published approximately 48% of them. America (all of North America + South America) published 11% of the articles and Asia published 30%. Finally, Africa and Oceania published the smallest number of articles, with respectively 7% and 4%.
Table 10 presents the countries that published more than two articles between 1990 and 2018. The empty cells in this table mean there is no publication which corresponds to the matching years and countries. This table also shows the increasing interest manifested in EA everywhere, with an accent in America and Europe. Particularly in the following countries: the United States, Iran, Australia, Sweden and the Netherlands, which published approximately 46% of the selected articles.
List of countries by publication occurrence.
When comparing these findings to the study area of the first authors, it shows that approximately 46% of the researchers who are studying in a Social Sciences area come from an academic institution located in Europe. In fact, European academic institutions seem to be showing more and more focus on this area of study.
What are the most common topics addressed in the articles?
The title of an article is the first clue to the topics addressed in this article. In order to have a broader view of the topics addressed in the selected articles, the most repetitive single words in their titles were used to create the word cloud presented in Figure 4. From ‘enterprise’ at 268 occurrences, ‘architecture’ at 214 occurrences, to ‘management’ at 22 occurrences, and ‘strategy’ with 4 occurrences, this word cloud supports the previous hypothesis concerning the increasing interest of Social Science departments in EA. Especially when observing how some words related to management, like ‘decision’, ‘structures’ and ‘strategy’ are more and more present in the titles of the articles.
Word cloud with the titles of the articles.
After reading and analysing the abstract, introduction and conclusion of the articles (at the very least), the following categories presented in Figure 5 were identified in accordance with the main topic addressed in each of them. EA tools: This category includes articles whose central aim is to study the tools developed for EA professional to achieve EA objectives, and the tools developed for an organization according to an EA approach. The particular contexts that compose this category are focused on descriptions, languages, patterns and architecture modelling. Some EA models and EA frameworks have also been developed or evaluated in this category. EA application: This category includes articles in which the central aim is to describe a specific use of EA which accomplishes a beneficial activity for the progress of an organization. It also includes articles whose objective is to provide a group of specific steps to follow when an EA strategy must be built, controlled and maintained. The particular contexts that compose this category are focused on the principles that guarantee a successful application of EA, the maturity of EA practice, findings of how to get the most value from EA and successful decision-making. EA discipline: This category includes articles whose central aim is to describe EA as a discipline and a practice in order to make its importance clear. In fact, the particular contexts that compose this category are focused on EA practice, challenges, roles, benefits and comparison to other fields. Some other articles of this category addressed the steps required to help EA become a recognized profession. In this category, many other publications have been reviewed to analyse and summarize the EA literature. The present article can be classified into this category. EA measurement: This category includes articles whose central aim is to evaluate and demonstrate the performance and maturity of EA. In fact, the particular contexts that compose this category are focused on aligning business and IT, compliance, return on investment and long-term financial improvement capabilities. EA practitioner: This category includes articles whose central aim is to highlight the mission and role of EA practitioners. The particular contexts that compose this category are focused on exploring the development and improvement of EA skills, and the strategies applied to achieve their mission.
Topics addressed in the articles.
This section shows how the EA community is focused on studying the development of new tools, and the optimization and analysis of existing tools (frameworks, models, etc.).
Qualitative findings
How do the articles approach EA?
Approximately 18% of the articles contain the term ‘enterprise architecture’ only in their title. Many of the articles explicitly used other terms to designate EA, like Information Technology, Information Systems Research, Organizational Modelling, Enterprise System Architecture, Architectural Approach and Enterprise Computing.
Many of the selected articles do not include any explicit or implicit EA definition. Researchers start talking directly about EA in these articles as if EA is a standard discipline, words or term that everyone is supposed to understand the established meaning of. Others of the selected articles do not provide personal definitions of EA but define it with one or several reference citations. Finally, just a few of the selected articles provide personal definitions of EA composed by the authors themselves, with their own words.
The significant importance of definitions in the identification of a discipline cannot ever be understated. In fact, the first question practitioners or researchers naturally ask whenever they engage with a subject for the first time is always: ‘what is this subject I am examining?’. 9 And the answer to such a question is a definition. Because of this, it is crucial to understand the meaning of EA from one article to another in order to allow people to be able to identify EA among other disciplines.
However, after reading the articles and looking at the associations they made with EA in their main sections, the following categories were extracted: Technological context (84%): The analysis, design, planning, implementation and other activities related to practicing EA are only focused on the ‘technological context’ of the organization. This category includes the conception of technological components, their evaluation, their alignment with the business and others. ‘This school is techno-economic in that it aims to reduce IT costs through technology reuse and eliminating duplicate functionality’.
4
Sociotechnological context (9%): The analysis, design, planning, implementation and other activities for conducting EA are not focused only on the ‘technological context’ of the organization but also on its ‘sociocultural context’. This category includes the management of people who are developing and using the technological components of the organization and their integration and participation in the decision-making process. Some references present this context as a top-down approach: ‘Traditional enterprise architectures are based on topdown approach. They emphasized on consistency throughout the organization and will involve all levels of employees’.
267
It is to say that ‘enterprise architecture is not only an IT issue, but a strategic and organizational challenge’.
268
Ecotechnological context (2%): The analysis, design, planning, implementation and other activities for conducting EA are not only focused on the ‘technological and social context’ of the organization but also on the ‘ecosystem context’. This category includes the relationships an organization has with its environment: other organizations, the community, the government, the environment, the ecosystem, the standards (requirements, specifications, guidelines, etc.) and so on. ‘Enterprise architecture should be able to cope with the fast changing business environment with ever changing needs and relations with the customer an boundaries’.
269
Five per cent of the articles were not considered for this classification because they too explore the lack of common understanding in the discipline of EA and present many similar ways of approaching the discipline without weight placed on one over another.
Table 11 presents an example corresponding to each category of focus. This does not imply that the authors of the cited references always work within the same context. The classification presented only corresponds to cited articles. Also, none of these three contexts should be considered above the others.
Examples of the focus of EA.
EA: enterprise architecture.
How do the articles approach the professionals who practice EA?
In addition to the previous observations concerning the context of EA on which the articles focused, they do not describe in the same way the role, mission, knowledge or competence of EA practitioners. In fact, in accordance with the different way to approach the practice of EA, as observed in the articles, the following categories were extracted: A ‘specialist’ or an ‘investigator’ who can imagine and understand the needs of an organization, the problems it is facing and the perspectives it is following in order to find and implement the best manner to satisfy or resolve them with IT. These enterprise architects think they can help organizations choose the best solutions to meet their needs.
233
An ‘integrator’ who has the ability to join all the stakeholders together with their understandings of the needs, perspectives and problems of their organization. These enterprise architects believe that IT alone cannot be an effective solution, but the participation and the motivation of the stakeholders in the decision-making process is crucial, and that effective solutions can be achieved through communication, negotiation and collaboration, for example.
83,217
A ‘facilitator’ capable of facilitating a good understanding of the needs of an organization, the problems it is facing and the perspectives it is following through the adaptation of these elements with the environment. Potential solutions must be adapted to the environment of the organization. These enterprise architects do not only focus on the internal environment of the organization, as the previous category does. Instead they believe that the organization can also be greatly impacted by the external environment (other organizations, the community, the government, the environment, the ecosystem, the standards, etc.), and vice versa. In fact, these enterprise architects think that IT and the social context of the stakeholders of the organization must also be accompanied by organizational adaptation to the outside world in order to take the lead in innovation and sustainability.
170
To what extent are the EA researchers/authors aware of the lack of common understanding?
As mentioned early in this article, there is an increasing number of authors who have described a lack of common understanding in EA. The analysis of the articles conducted in this study reveals that many of these authors are aware of a challenge caused by the existence of different, and even divergent, understandings of EA. One author explains, for example, that ‘EA is still a challenging concept’ because there is no universal world view in EA, but several definitions of EA exist and there are various perceptions. 161 Another explains ‘EA lacks semantics’, and that people cannot have an exact and common understanding of EA. 186
Some other articles are more to-the-point and affirm for example that EA suffers from ambiguous definitions of what it is or is supposed to be. Another highlights ‘an absence of any consensus’ concerning what EA is or supposed to do and how it is supposed function. 185 Yet another indicates ‘a lack of theoretical foundation, definition, or common understanding’ among researchers who have published in EA. 270 Still others address this issue by questioning the differences between the approaches of enterprise architects. For example, 99 explains how there are an increasing number of enterprise architects, ‘but there is no universally accepted baseline of standards and knowledge to ensure consistent service’. And 225 explains how variation and contradiction identified in the EA definitions within the literature ‘further complicates the challenges of defining the role’ of EA practitioners.
Despite this increasing number of authors who have reported a lack of common understanding in the discipline of EA, few of them proposed to fully investigate, understand or resolve this challenge. However, certain studies try to generate new ways of approaching EA based on several existing definitions and concepts of EA. Certain other studies try to demonstrate how some ways of approaching EA correspond or not to the practice of EA.
Finally, another significant consideration that the articles analysed in this study reveals is the consequences of the lack of common understanding in the value of EA. Is it clear that the use and usability of EA may fully depend on ‘how it is understood, defined and scoped’. 102 In effect, without the presence of concise and precise description concerning the roles that can achieve architecture success, ‘architects may be viewed as providing no specific value’ for organizations. 133
Discussions
Discussions concerning the findings
Concerning the distribution of the articles over the years, the articles selected for this mapping study do not represent the total number of journal articles published in EA from 1990 to 2018. This is because of the limitation of the inclusion criteria applied, the duplication of some of the articles and the articles that are non-downloadable. Moreover, this study includes only a portion of the articles published in 2018 because the search was conducted in mid-2018. But comparisons with the articles selected in some SLR concerning a general summarizing of EA 13 – there are no other SMSs concerning a general summarizing of EA literature to be considered – show that a large majority of the published journal articles were considered in this study and then the sample is representative of the total number of publications (population).
Taking into account the previous precision, observing the distribution of the articles over the years provides useful insight into how young EA still is. For example, the highest number of journal articles published in 2014 is 33. Without a doubt, this number is small compared to the number of published articles in the field of Software Engineering for example, which is also a recent discipline of study. This argument is not intended to declare that the discipline of EA is not generated growing interest. In contrast, as indicated in the beginning of this study, the growing number of EA publications over the years, the growing number of practitioners and researchers involved in EA research and the growing number of conferences and training organized for EA are a perceptible proof of it evolution. The various topics that have been developed in EA literature and the diverse approaches and techniques that have been used to investigate these topics can also be considered as a concrete sign of the evolution of EA.
Concerning the experiences of EA researchers/authors who have published in EA (RQ1), when analysing the fact that approximately 65% of authors included have published only one of the articles, it seems that a large majority of the authors of EA literature are not experienced researchers in EA. This leads us to ask why EA researchers do not become mainly focused on EA? Are there some EA researchers/authors who mainly work on EA as their area of specialization? Do EA researchers/authors consider EA as a sub-branch of other main disciplines or as a separate branch derived from other disciplines?
Concerning the occupation of the authors (RQ2), they are predominantly students/researchers and professors/researchers, because a large majority of them are affiliated with an academic institution. A specific restriction in academic research is that new observations and argument must regularly derive from existing references. Because of this obligation, maybe there would not be so many ways of approaching EA in the literature if EA authors had agreed references to follow. This raises numerous questions, such as: Do EA researchers/authors have agreed and standard references to follow, including for example definition, terminology and world view? Why have academic authors/researchers have so many ways to approach the discipline of EA? It would be interesting to know how many of the articles are written by students/researchers with their supervisors, and how many are written only by professors/researchers, in order to evaluate which of these two scenarios present more variations (i.e. definition, terminology and world view) compared to existing references.
Concerning the academic disciplines in which researchers have studied (RQ3), at least three categories – Information Technology, specific areas of engineering and social and human sciences – were found. Undoubtedly, each of the fields from which the discipline of EA has originated has a different world view including different ways of perceiving and facing real-world problems and procuring results. What is the impact of the world view of each of these fields on the final approach that authors provide to EA?
Concerning the focus of the publishers/editors of the EA publications (RQ4), there is an absence of enough journals and editors/publishers dedicated specifically to EA. In fact, the institutions which have published the most articles are the well know publishers that often have disciplines related to IT as main subject areas. Because there are not enough publishers dedicated specifically to EA, the articles are also published here and there through various journals.
The analysis of the subject area of the institutions which have published the EA papers also shows how the Social Sciences are more and more represented in EA even though a large part of the research is conducted by researchers that have studied in IT and an Engineering area, and published by editors/publishers with a subject area and category related to the same disciplines.
Concerning the location of the first author’s affiliated organization (RQ5), English is only the official language of 38% of the countries where the affiliated organizations of the first author are located, while only articles written in English were selected in this study. Because of this, would it be reasonable to consider sufficient knowledge of the English language to also be a factor favouring the existence of different ways to approach EA in the literature? Furthermore, it would be necessary to confirm the authors’ languages in order to support such a hypothesis.
Despite the fact that 17% of these articles are written by first authors from the United States, only 11% of these articles are from the American continent. In fact, European researchers/authors – 48% of the articles are written by first authors from Europe – seem to have taken control of the leadership of the EA discipline. 7
Approximately 47% of the researchers who are studying in a Social Sciences area come from a European academic institution. When observing that the majority (60%) of the articles with unknown study areas of the first author (absent 17%) also come from the European continent, it is possible to imagine that the authors of these articles are also studying in Social Sciences. If so, this will increase this category of authors who are studying in Social Science (14%) which is actually lower than the authors who are studying in IT and an Engineering area (54%). This supports the previous observation which indicated that the social and human sciences are more and more represented in EA. The word cloud shown in Figure 4 (RQ6) is further evidence which supports that the managerial context of the organization is more and more considered in EA research, even when the technological context is dominant. In effect, this aspect can be observed in the increasing use of certain words even in the titles of the articles which explicitly refer to social and human sciences.
Another aspect concerning the most common topics addressed in the articles (RQ6) concerns how the evaluation of the utilization of EA tools, either newly developed or previously existing, have been neglected in the literature of EA. It seems that there is a lack of relevant directions for future studies in EA. In effect, the majority of the publications are focused on building and studying EA tools developed to apply EA or tools derived from an EA application (EA tools 55%). But without a complete and up-to-date understanding of the practice of EA (EA practitioner 4%) – including the role of EA practitioners, their world views and their needs, for example – How will it be possible to create appropriate tools for them? Without clear evaluation (EA measurement 7%) of the performance of the existing tools – including the characteristics to measure and their importance, the metrics and the standards, for example – How will it be possible to continually improve their creation and use? Conducting more literature analyses (i.e. SLRs, SMSs and content analyses) intended to study the state of the art of EA or to explore specific challenges concerning EA could help provide relevant directions for future studies. For example, this could help researchers to avoid fundamental work on EA tools when several existing tools have not been applied (EA application 23%) or evaluated yet. In effect, the practical aspect of EA must also play a more important role in EA research through the realization of more descriptive and experimental research which uses explicitly corresponding research methods such as opinion surveys, discourse analysis, participatory action research and design science research, for example.
Concerning the ways that the articles approach EA (RQ7), the original data collected without any interpretation prove the existence of the lack of common understanding in EA. The various definitions provided to explain what EA is, what value EA is supposed to provide organizations, how EA is supposed to be applied and the various other terms used to designate EA are some examples. The indication of this lack of common understanding in EA in more and more articles, as seen in the findings, has demonstrated how EA researchers/authors are aware of this lack (RQ9). Now, this challenge must be studied in depth in order to find more tangible findings that can help to better address it. The characteristics and assumptions discussed in the previous sections represent precisely some important characteristics which can be taken into account in order to study this lack of common understanding. Answers to the different questions generated would be very useful for a better understanding of the origins of this lack. However, these characteristics – complemented by others – are not required to be analysed individually. Many other questions must be asked in order to relate them, and many other questions must be asked concerning the methodological techniques that will allow us to find the appropriate answers. For example, the fact that more publications are focused on EA tools can be caused by the choice of the publishers to publish mainly articles in this category rather than the others. Just as it can be caused by the academic discipline in which EA practitioners have studied.
The categories found concerning how the articles approach EA (RQ7) which are the ‘three major ways of approaching EA’ (technological, sociotechnological and ecotechnological) are based the ‘three modes of EA’, 208 the ‘three schools of thought on EA’ 4 and the ‘three distinct interlinked architectures’. 5 The difference in this study is that each of these categories is presented only according to the information extracted from the articles (contexts of the focus and the tasks). This means that other interpretation did not take place in order to provide a full description of each category (scope, assumption limit, etc.). At the first observation, it seems that the way of approaching EA is strongly connected to the discipline in which the first author has studied (technological context → IT areas; sociotechnological context → engineering areas; ecotechnological context → social and human sciences). But the findings do not confirm such an assumption because an overwhelming majority of articles correspond to the technological context.
On the other hand, the three ways the articles approach professionals practicing EA (specialist, integrator and facilitator) (RQ8) derive from the previous ways of approaching EA. Because a large portion of the articles focused on building, they have presented EA practitioners as specialists who can create, modify and optimize (i.e. tools, processes, principles, documentations and strategies) without involving all the stakeholders in the decision-making process to be sure to understand their needs and motivations (internal environment), as well as the interest of the whole community (external environment).
Building a codebook – including the specific words, expressions and wording – which identifies the particularity of the articles placed in each of the ways of approaching EA and its practitioners could be an appropriate method (content analysis) to validate these findings.
Implications for research
A large number of studies reported that many ways of approaching EA exist, even if it is not their main focus. In fact, only a few studies are completely dedicated to investigating this lack of common understanding in EA. Until this moment, the studies which are completely dedicated to studying this lack of common understanding in EA do not use a rigorous investigation and thus based their findings on primary studies selected and analysed without following specific criteria. Therefore, a survey was also conducted on this topic. But this survey used the existing models and did not leave enough opportunity to draw a complete picture of the state of the art of EA.
The situation described above shows that validity and reliability are mostly missing in the investigations which address lack of common understanding in EA, and also more investigations must be conducted. In this context, the contribution of this study is manifold.
First, it represents one of the few studies which address this problem of lack of common understanding. It confirms some previous findings and provides new insights which can be taken into account for future studies on the same and corresponding topics.
Second, compared to the few previous studies on this topic, this is the first one which analyses the literature with rigour in accordance with the guidelines of the well-known scientific method, which is SMS. This allows this study to show greater validity and reliability that researchers should consider going further. This study also provided significant insights for future research on the same topic. In fact, within the findings or even in the discussions, many new considerations which require deeper investigations were made. For example, the experiences of EA researchers, and the impact of the authors’ first languages, or the discipline on which they studied, on the lack of terminology.
Third, compared to the few previous SMS in the discipline of EA concerning other topics, this study provides some new observations that can complete the existing state of the art of EA as described in the literature. For example, no previous SMS on EA has focused on the number of articles published by each EA author/researcher, the academic disciplines in which they have studied, or their occupation when publishing. No previous SMS on EA had focused on the subject areas of the publishers of the EA publications or on the occurrences of certain words in the publication titles. But even the importance of such subjects in the context of this study, as it can be seen the findings concerning them should also be considered to show a complete presentation of the state of the art of EA. Researchers could also use this information as a starting point to summarize the EA literature with all the important details.
Fourth, compared to many previous SMSs which their predefined classification schemes in advance, this study has generated categories which emerged progressively during the data extraction. This method provides better opportunity to summarize the entire content of the sources analysed without losing the details.
Implications for practice
The lack of common understanding in EA can create misunderstandings and conflicts regarding the role and responsibility of professionals practicing EA. Especially when EA team members are not thoroughly conscious of the lack of common understanding in EA and the extent of the existing differences. It can also be hard to collaborate with stakeholders and other participants in such situations. Similarly, it can be hard to provide standard and universal training to future EA practitioners. And researchers can face difficulty when sharing their research findings and generally being understood.
The previous studies concerning the lack of common understanding have presented the most popular schools of thought on EA, while the current study has focused on the extraction of the details which can help to differentiate and link these schools of thought. This means that the information collected and analysed in this study is at a lower level and thus can be more meaningful for practitioners. In fact, this study is useful to help professionals practicing EA to be conscious of the existence of many different contexts, which could otherwise prevent EA professionals from having common terminology, understanding and perspective. This study could open many ways to help them become more tolerant of each other and collaborate better.
Taking into account the consideration of the previous sections, it is evident that this study could also help the administration staff of the organization to better know the kinds of EA professionals they need, depending on what the organization want to achieve. This study could also help human resources to be better able to evaluate candidates according to the need of the organization. In the same line of thinking, this study could motivate the integration of all the existing perspectives in the EA academic programmes, in order to provide universal training to future practitioners.
However, one point to be clarified is the importance of each of the ways of approaching EA, without any superiority of one over another, even if they seem to be divergent and conflictual sometimes. The objective is to understand the underlying assumptions of the different perspectives, beliefs and world views underlying the many ways of approaching EA and its practitioners in order to integrate them all into a shared reference. This will allow us to take them all into account when conducting research, elaborating tools, organizing training, creating job offers, implementing EA plans, projects or processes and more. In effect, this will allow enterprise architects and researchers to better collaborate even if they have different ways of approaching EA.
Conclusion and future work
This study conducted a SLR and analysed 257 journal articles published from 1990 to mid-2018 with the aim to identify, explore and classify elements that might influence the existing lack of common understanding in EA. The findings confirm that the extent to which the authors/researchers are focused on EA, the sectors in which they are evolving, the academic disciplines in which they have studied, the countries where their affiliated organizations are located, the subject areas of the journals/publishers of their publications and the way they have approached EA and its practitioners were identified as sources of variety which could be at the basis of the existing lack of common understanding in EA.
A limitation to note is that this study analysed only journal articles in order to keep it to a manageable size. Despite this limitation, the contribution of this study – which is the first SMS on the lack of common understanding in EA – is the organization of the EA literature according to three major questions concerning ‘who’ has been published in the literature, ‘where’ they have been located and ‘what’ their publications are about. This helps to better identify sources of variety which could be on the basis of the lack of common understanding in EA and provides practitioners and stakeholders a better understanding of this challenge. This also provides relevant directions for future studies. Due to this limitation, future studies on this topic must include other relevant data sources, such as conference articles, book chapters and more, and use other reliable methods, such as SLRs, content analyses, surveys and case studies.
Footnotes
Authors’ note
This article is the extended version of a previous conference article. The contribution of the previous article was extended both in content and in depth. The following sections present some of the relevant modifications added in order to improve this study. Firstly, the findings of this article take into consideration the enterprise architecture (EA) journal articles published from 1990 to mid-2018, while the previous study only considered the publications from 1990 to 2014. Consequently, 257 articles were analysed in this study, while 171 articles were analysed in the previous study. To achieve this, the search string ‘enterprise architectures’ (plural form) was also added to those considered in the previous study (EA, enterprise architect, or enterprise architects). Secondly, the section of research design was more detailed and includes a section for each step of the systematic mapping study applied, according to up-to-date references. Thirdly, the section of findings has been revised with the results of the new articles added and has been completed with more detailed explanations, tables and figures. Additional analysis and related findings with regards to the intensity of publications of authors that have published in EA, the subject areas and categories of the publishers/editors of the EA publications and how the articles approach EA and EA practitioners were also presented. Fourthly, the section of discussions was rewritten in a more systematic way in order to identify the discussions for each finding and to provide several critical questions which can provide relevant directions for future study concerning the existing lack of common understanding in the discipline of EA. Two sections which describe the implication of the study for researchers and practitioners were added.
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) received no financial support for the research, authorship, and/or publication of this article.
