Understanding Enterprise Architecture with Topic Modeling

Preliminary Research based on Journal Articles

Marco Nardello

1,*

, Charles Møller

and John Gøtze

2,3

Department of Materials and Production, Aalborg University, Aalborg, Denmark

IT University of Copenhagen, Copenhagen, Denmark

QualiWare aps, Farum, Denmark

Keywords: Enterprise Architecture, Topic Modelling, Content Analysis, Research Theme, Future Research, Machine

Learning, Latent Dirichlet Allocation.

Abstract: The next 3 years will be more important than the last 50 due to the digital transformation across industries.

Enterprise Architecture (EA), the discipline that should lead enterprise responses to disruptive forces, is far

from ready to drive the next wave of change. The state of the art in the discipline is not clear and the

understanding among researchers and practitioners is not aligned. To address these problems, we developed

a topic model to help structure the field and enable EA to evolve coherently. In this preliminary study, we

present the 360 identified topics in EA literature and their evolution over time. Our study supports and

combines the findings from previous research and provides both a deeper analysis and more detailed findings.

1 INTRODUCTION

Enterprise Architecture (EA) leads enterprise

responses to disruptive forces (Gartner Inc. 2017) and

industries are now on the brink of digital

transformation. Globally 72 percent of CEOs believes

that “the next 3 years will be more critical for their

industry than the last 50 years” (KPMG 2016, p.8).

KPMG also explained that “the speed of change will

be exponential” and 77 percent of CEOs “are

concerned about whether their organization is

keeping up with new technologies” (KPMG 2016,

p.7). However, the current state of the art of EA is not

clear, and its fundamental concepts are not shared

among researchers and practitioners (Saint-Louis &

Lapalme 2016; Rahimi et al. 2017). In addition, from

informal interviews the authors acknowledged that

EA researchers and practitioners find problematic to

position themselves in the field because they are

lacking its overview.

Two studies tried to address this problem

systematically (Simon et al. 2013; Saint-Louis &

Lapalme 2016) though their limited scope left

important research themes uncovered. Simon et al.

(2013) conducted a content analysis of EA including

articles published until 2010. Since then EA literature

has almost tripled (Saint-Louis 2016). A more recent

example, the Systematic Mapping Study (SMS) of

Saint-Louis and Lapalme (2016) focused on literature

from selected journals therefore excluding most

literature on EA. In addition, based on Debortoli’s et

al. (2016) article both studies are subject to human

bias. In the first, the authors conceptualized EA in a

model and classified literature applying it (Debortoli

et al. 2016). In the second one, the authors manually

identified topics.

These studies exclusively identified the topics

without specifying the concepts related to them and

their evolution over time. To date there is no study

that identifies topics in EA without major human

subjectivity bias. In addition, directions for future

research are mainly based on personal experience

(Korhonen et al. 2016; Lapalme et al. 2016).

Another problem, not restricted to EA, is the lack of

a structure of research fields and how they evolved

over time.

To identify EA’s topics and their evolution over

time we developed a Topic Model applying Latent

Dirichlet Allocation (LDA) method by Blei et al.

(2003). This method develops a list of topics, for each

one presents the most frequently occurring words and

identifies the most relevant literature (Blei et al.

2003). The LDA method has been applied

successfully in other fields like in Information

Systems research (Chen & Zhao 2015). In this

preliminary study, we investigate the following

640

Nardello, M., Møller, C. and Gøtze, J.

Understanding Enter prise Architecture with Topic Modeling.

DOI: 10.5220/0006693106400648

In Proceedings of the 20th International Conference on Enterprise Information Systems (ICEIS 2018), pages 640-648

ISBN: 978-989-758-298-1

research questions:

1. What are the topics in EA literature?

2. How did topics in EA literature evolve over

time?

We present a topic model of EA based on 602 journal

articles. An article following this preliminary study

will present a topic model including also non-journal

publications.

This study aims to guide coherently the

development of EA research and it mainly contributes

to the EA community in two ways. First, it allows the

community to identify topics in EA that are relevant

for them, opening new research opportunities.

Second, it identifies the literature related to each

topic.

A final contribution to research in general is to

present a new approach for structuring research fields

through the development of a topic model. This type

of efforts can support systematic literature reviews by

structuring the field and providing the context and

literature for this type of research.

The article continues presenting the background

literature in section 2, the method applied in section 3

and the key findings of this study in section 4. We

discuss the results in section 5 and the conclusions

and future research are presented in section 6.

2 BACKGROUND

The foundations of EA date back to the 1960s when

IBM initiated the Business Systems Planning (BSP)

methodology (Kotusev 2016). BSP was “a structured

approach to assist a business in establishing an

information systems plan to satisfy its near- and long-

term information needs” (IBM 1978). In the 1980s,

EA emerged from BSP (Kotusev 2016). Zachman

(1987) developed a framework for information

systems and Spewak and Hill (1993) defined

Enterprise Architecture Planning. EA Planning was

“the process of defining architectures for the use of

information in support of the business and the plan for

implementing those architectures” (Spewak & Hill

1993). Finally at the beginning of the new

millennium, The Open Group Architecture

Framework (Group 2009) developed a new EA

reference architecture and methodology that are today

widely used in practice (Simon et al. 2013).

Two articles mapped EA contributions (Simon et

al. 2013; Saint-Louis & Lapalme 2016). The first one

is a combined bibliometric study and content analysis

of EA research from Simon et al. (2013). In their

study, the top down content analysis classified EA

publications in a predefined scheme. In this way, the

authors biased the results of the content analysis by

defining the classification scheme based on their

understanding prior the study (Debortoli et al. 2016,

pp.112–113). In addition, analysing Saint-Louis' and

Lapalme's EA article distribution by year (2013, p.76)

is possible to infer that Simon et al. (2013) study did

not consider almost two thirds of EA literature

currently available. This is also supported by a search

on Scopus database for articles with "Enterprise

Architecture" in title, abstract and keywords. Around

1300 articles were published before 2010 and almost

2000 were published after.

The study of Saint-Louis and Lapalme (2016)

reviewed 171 journal articles. They presented a

bibliometric study and a summary of ten

conceptualizations of EA, namely framework, model,

discipline, integration, measurement, strategy,

principles, design, literature, and practitioner. One

limitation of their study is to have included only

journal articles from few sources. Other sources were

not considered. In addition, when the authors

classified EA topics they assigned only one topic per

article. This is restrictive because it is common for a

research article to include multiple topics (e.g., an

article about EA models for business alignment

covers both the EA model topic and business

alignment topic) (Chen & Zhao 2015, p.2).

Another challenge in EA research is that efforts

are weakly integrated (Simon et al. 2013, p.19). This

is supported by the fact that systematic studies in EA

are very limited in scope. Excluding the SMS

previously mentioned, the other three available in the

field focus respectively on enterprise integration

(Banaeianjahromi & Smolander 2016), applications

of ontologies (Pinto et al. 2014), measurement and

indicators (Abdallah et al. 2016). Among the

systematic literature reviews, we found areas as the

automated analysis and documentation (Florez et al.

2016; Farwick et al. 2016; Hauder et al. 2012),

aspects of EA Management (Kotusev 2017; Rahimi

et al. 2017; Jugel et al. 2016; Huber et al. 2017;

Brosius 2016; Lange et al. 2016; Schneider et al.

2013; Wißotzki & Sonnenberger 2012) and a research

group involved in EA implementation (Rouhani et al.

2015; Nikpay, R. Ahmad, et al. 2017; Nikpay, R. B.

Ahmad, et al. 2017).

Our study aims at overcoming the limitations of

previous research, outline the topics in EA research,

and present their evolution over time.

Understanding Enterprise Architecture with Topic Modeling

641

3 METHODOLOGY

There are five main methods for performing text-

categorization (Debortoli et al. 2016) – bottom-up

and top-down manual coding, dictionaries, and

supervised and unsupervised machine learning. We

applied the latter that uses documents to inductively

discover categories and assigns documents to these

categories (Debortoli et al. 2016). We chose this

method because it “generates reproducible results that

are not subject to the human subjectivity bias”

(Debortoli et al. 2016).

An application of unsupervised machine learning

is probabilistic topic modelling. This approach is

based on the assumption that “words that occur in the

same contexts tend to have similar meanings” (e.g.

the co-occurring words “mozzarella”, “tomato”,

“basil”, “margherita”, “oven” all refer to the “pizza”

topic) (Turney & Pantel 2010, p.142). Three main

methods are used in Information System research

(Debortoli et al. 2016): Latent Semantic Analysis

(LSA), Latent Dirichlet Allocation (LDA) and

Leximancer. We applied LDA because previous

studies proved that LSA methods suffer from

interpretability issues (Debortoli et al. 2016) and

Leximancer algorithms are scarcely documented

(Debortoli et al. 2016).

LDA methods are grounded on the assumption that

documents are generated from a set of topics and that

each topic is characterized by a distribution over

words (Chen & Zhao 2015, p.2). For this reason, a

document can contain multiple topics. In this way, we

identified research topics in EA discipline.

In this study, for the data gathering and data

filtering steps we followed (Petersen et al. 2015), and

for the data preparation, application of text-mining

techniques, and evaluation of the topic model we

followed (Boyd-Graber et al. 2014; Debortoli et al.

2016). The first three steps have been performed

manually while the last two were automated.

3.1 Data Gathering

We included journal articles that contained the words

“Enterprise Architecture” in the title, abstract or list

of keywords. The search was performed in July 2017

on the following databases (in parenthesis the number

of articles retrieved): Scopus (539 articles),

ABI/INFORM (167 articles), Business Source

Premier of EBSCO (205 articles), Web of Science

(287 articles), Compendex (306 articles), INSPEC (0

articles), IEEE Xplore Digital Library (31 articles)

and AIS Electronic Library (61 articles). The search

generated 715 unique articles. Depending on the

database the search was restricted to journal articles

or to peer-reviewed articles. When possible, we

limited our search to articles written in English. We

decided to include only journal articles to focus this

preliminary study on what we consider the most

reliable outlet of contributions in the field.

3.2 Data Filtering

Each combination of title and abstract represented

what we will refer to as a document. The first author

performed this filtering alone since in case of a large

number of studies and many of them are clearly

identifiable noise the process may be conducted

individually (Petticrew & Roberts 2008). When in

doubt, he was inclusive. The following criteria were

applied based on (Petersen et al. 2015). Inclusion

criteria:

The study is published in a journal

The study relates to EA, as defined in the

Introduction section.

Exclusion criteria:

Conference or journal editorials

Studies presenting book reviews

Studies not written in English

Books and grey literature

The filtering process had two steps, see Figure 1.

First, we removed all the articles that were not journal

publications. Second, we removed articles not related

to EA based on title and abstract. We used the

resulting 602 articles as input for the topic model.

Figure 1. Filtering process.

3.3 Data Preparation

When preparing the documents for the analysis we

followed the guidelines from Boyd-Graber et al.

(2014). We started by removing punctuation

notations from the documents (except for the dash

that connects words closely related) and formatted the

text in small caps. We did not apply stemming since

ICEIS 2018 - 20th International Conference on Enterprise Information Systems

642

it combines terms that may have different meanings.

Next, we used Blei’s and Lafferty’s Turtotopics to

identify multi-word expressions (Blei & Lafferty

2009). They are also known as n-gram and “business

IT alignment” is an example of 3-gram word that

captures a single concept in EA. We used for this

analysis only the abstracts with the following settings:

p-value 0.1 and the n-word appears at least 3 times in

the entire corpus. Since some phrases were repeated

in shorter forms, we considered only the shortest form

that preserved the meaning of the longer phrases (e.g.

we considered “enterprise architecture framework”

instead of “enterprise architecture framework eaf”).

We modified the words in the document to

distinguish them from 1-gram words (e.g.

“EnterpriseArchitectureFramework”).

Afterwards, we extended Stanford’s Topic

Modeling Toolkit list of English language stopwords

(Ramage & Rosen 2011) with common words

pertaining to EA literature (see the Appendix for the

full list). We used this list to exclude all the non-value

adding words from the analysis. Removing the n top

words is an alternative approach, thought the result

were worse than using the stopword list.

3.4 Text-mining

For the text-mining we used Stanford Topic Modeling

Toolbox (v0.4.0) (Ramage & Rosen 2011). We

computed the topic model with different number of

topics 10-fold ranging from 10 to 450. The LDA

method had the following parameters: max iteration

3000, topic smoothing 0.01, term smoothing 0.01. For

training the model to fit the documents we

experimented with the two main approaches – Gibbs

sampler and variational Bayes approximation – and

decided to use the latter since it produced better

results. At 360 topics the results were most

understandable. We did the topic labelling manually.

3.5 Evaluation

To evaluate the number of topics, we calculated the

perplexity of held-out documents with different

number of topics from 10 to 450 in 10-fold.

“Perplexity is a standard performance measure of

different models in natural language processing; a

lower value of perplexity indicates a better model

performance” (Chen & Zhao 2015). As the number of

topic increases, the perplexity decreases until 360

topics and after that it stabilizes.

4 FINDINGS

In this study, we identified 360 topics in EA research,

the keywords related to each of them and their

evolution overtime. Due to space limitations, we will

present only the ten topics that covered most of EA

body of knowledge. For these topics, the five most

recurring words and the percentage of coverage of the

topic over the literature are available in the appendix.

The full list of the topics is available online

(QualiWare 2017).

First, the Business Process topic refers to business

processes (e.g. in the financial industry) and includes

concepts and information to support decision makers.

Second, the Framework topic relates to EA

frameworks that are adequate for integrating and

structuring information. They can be used to capture

the baseline structure of the organization as well as to

clarify and achieve predetermined outcomes.

Third, the Enterprise Resource Planning (ERP)

Model topic refers to models of companies ERP

software specifying its requirements, features and

functions. It is related to the Business Process topic

and to the Business Process Reengineering practice.

Fourth, the Management Activity topic presents

activities done by managers that include objectives,

opportunities, clients, necessary resources,

information, and information security. These

activities relate also to the concept of holistic

approach.

Fifth, the Design topic refers to the design of the

components and their interaction of Business

Processes, Systems, Information Technology (IT) in

order to support the analysis and the collection of

information.

Sixth, the Modelling Methodology topic refers to

methodologies and approaches for modelling

efficiently and systematically different facets of the

organization. Two main fields of application are the

healthcare sector and the Virtual Enterprise (VE).

Seventh, the Analysis Method topic includes the

concepts of techniques, estimations, accuracy,

uncertainty, and validation. This topic is scientifically

investigated mostly through case studies.

Eight, the Meta Model topic is the one that covers

the most body of knowledge. It related to meta-

models and the Analysis Method and Models topics.

Ninth, the Model topic relates to integrated models

that support the evolution of company’s systems.

Finally, the Service topic refers to services and

includes concepts like users, resources, infrastructure,

and information systems.

For each topic, we plotted its usage over time by

using Stanford Topic Modeling Toolbox feature

Understanding Enterprise Architecture with Topic Modeling

643

Figure 2. Topic distribution over time.

“slice”. This feature creates subsets of the results

associating it with one or more variables (e.g. year of

publication, source, authors, and so on). By splitting

the results for each year it is possible to understand

when each topic emerged and which have a growing,

stable or decreasing trend. Figure 2 presents the

evolution of the topics previously described (the

vertical axis represents the total number of words

associated with each topic for each year).

Four topics emerged in the last century. The first

topics that appeared in EA research are the

Framework, Modelling Methodology, Analysis

Method and Model. The Framework topic was

introduced in 1990 and has it started being

significantly researched from 2005 and it reached its

peak in 2014. Research on the Modelling

Methodology topic was popular around the year 2000

and since then has been less researched. The Analysis

Method topic emerged in 1998 and it had a constant

growth from 2008 until it reached its peak in 2013.

Since then its has been less researched. Research on

the Model topic started in 1990 and grew constantly

from 2004 until now.

The remaining six topics were researched from

the year 2000 onwards. Starting with the Business

Process topic, it grew constantly from the year 2000

until 2010 and since then it became less researched.

The ERP model topic emerged in 2001 and peaked in

2012 and 2013. Since then research on this topic

decreased significantly. The Management Activity

topic started to be more researched in 2004 and since

then had a fluctuating trend. The Design topic

emerged in the year 2000, it peaked in 2006 and since

then had a decreasing importance. The Meta Model

topic emerged in 2004 and peaked in 2010. Since then

it had a fluctuating trend. Finally, the Service topic

grew constantly from 2003 until 2008. Since then it

had decreasingly researched.

Topic models also provide a new way of

navigating literature. As an example of topic-based

literature navigation we list the three articles that are

contributing the most to the Modelling Methodology

topic. First, Glazner (2011) applied simulation

methods in EA. Second, Kim et al. (2006) developed

a systematic modelling approach for VEs. Third,

Nugraha et al. (2017) presented “a business

architecture modeling methodology to support the

integration of primary health care”.

5 DISCUSSION

In this section, we will present the implications or our

research and its limitations.

The three main implications of our study are to

provide a better understanding of EA topics, integrate

existing findings, identify new topics.

ICEIS 2018 - 20th International Conference on Enterprise Information Systems

644

First, our study helps researchers and practitioners

understand the topics in EA. It does so in three ways,

identifying the concepts related to each topic, in

which fields it has been mostly researched, and the

literature related to it. For instance, the Modelling

Methodology topic refers to modelling various facets

of the organization and not exclusively the IT or

business aspects. In addition, the focus of research

has been on efficient and systematic approaches. Our

topic model helps researchers and practitioners

working with Modelling Methodologies to

acknowledge of these concepts and the focus of

research. Afterwards, the Modelling Methodology

topic helps researchers and practitioners understand

in which fields this topic is most applied, health care

sector and VE. Finally, our study identifies the

articles related to the topic.

Moving on to the comparison of EA topics with

previous studies, we will use the top ten topics to

illustrate how the topics identified in this study are

related to previous studies. The details of this analysis

are available in the appendix. Seven topics are

covered in the content analysis of Simon et al. (2013)

and six are related to the topics identified by Saint-

Louis and Lapalme (2016). The topics in common are

on a higher abstraction level than the others. The two

that are not mentioned in previous studies are the ERP

Model and Service topics. In this study, depending on

the abstraction level of the topic some of them are

inline and combine the findings of previous studies

while others present new, and usually more specific,

research topics.

This latter point combined with the topic-guided

literature navigation can support literature studies by

making available articles that might be left out from

a keyword search.

Our study has three main limitations. First, the

decision to include only journal articles might have

caused significant contributions to be left out from

this study. Having said so, this study is with Simon et

al. (2013) among the ones that included the widest

body of literature. Second, concerning the topics and

their labelling, the model has been evaluated only by

the authors. Even though both co-authors publish in

the field of EA, external evaluation would improve

the reliability of our results. A third limitation is to

have analysed only the abstracts of the journal articles

and not the full text. Having said so, this is a standard

approach in Topic Modelling (Chen & Zhao 2015).

6 CONCLUSION

In this study, is a first step towards a map of topics of

EA. This preliminary research identified topics in EA

research based on 602 journal articles. We presented

the ten topics that cover most literature in the field

and based on these we discussed the contribution of

topic modelling to EA research. We presented how

these topics improve the understanding of EA

literature, how they integrate and combine previous

findings and how they can be used to navigate EA

literature with a topic-guided approach. In addition,

we have discussed how topic models can support

systematic literature reviews in EA.

Future research will focus on extending the body

of literature analysed to include also non-journal

publications. In addition, different text mining

approaches will be investigated to enhance the value

of topic models in EA research. Interesting

contributions that might be applied and further

developed are the hierarchical topic modelling (Blei

et al. 2010), correlated topic modelling (Blei &

Lafferty 2007), automatic labelling technique (Lau et

al. 2011) and supervised machine learning.

REFERENCES

Abdallah, A., Lapalme, J. & Abran, A., 2016. Enterprise

Architecture Measurement: A Systematic Mapping

Study. 2016 4th International Conference on

Enterprise Systems (ES), pp.13–20. Available at:

http://ieeexplore.ieee.org/document/7880427/.

Banaeianjahromi, N. & Smolander, K., 2016. What do we

know about the role of enterprise architecture in

enterprise integration? A systematic mapping study.

Journal of Enterprise Information Management, 29(1).

Blei, D.M., Griffiths, T.L. & Jordan, M.I., 2010. The nested

chinese restaurant process and bayesian nonparametric

inference of topic hierarchies. Journal of the ACM

(JACM), 57(2), p.7.

Blei, D.M. & Lafferty, J.D., 2007. A Correlated Topic

Model of Science. The Annals of Applied Statistics,

1(1), pp.17–35. Available at:

http://www.jstor.org/stable/4537420.

Blei, D.M. & Lafferty, J.D., 2009. Visualizing topics with

multi-word expressions. arXiv preprint

arXiv:0907.1013.

Blei, D.M., Ng, A.Y. & Jordan, M.I., 2003. Latent dirichlet

allocation. Journal of machine Learning research,

3(Jan), pp.993–1022.

Boyd-Graber, J., Mimno, D. & Newman, D., 2014. Care

and feeding of topic models: Problems, diagnostics, and

improvements. Handbook of Mixed Membership

Models and Its Applications, pp.225–254.

Brosius, M., 2016. Motivation for Coordination-a

Complementary Approach to Enterprise Architecture

Management Research. In Enterprise Distributed

Object Computing Workshop (EDOCW), 2016 IEEE

20th International. IEEE, pp. 1–8.

Understanding Enterprise Architecture with Topic Modeling

645

Chen, H. & Zhao, J.L., 2015. ISTopic: Understanding

information systems research through topic models.

Icis-Rp, pp.1–10. Available at:

http://aisel.aisnet.org/icis2015/proceedings/GeneralIS/

6/.

Debortoli, S. et al., 2016. Text Mining For Information

Systems Researchers: An Annotated Topic Modeling

Tutorial. Communications of the Association for

Information Systems, 39, pp.110–135. Available at:

http://aisel.aisnet.org/cais%0Ahttp://aisel.aisnet.org/ca

is/vol39/iss1/7%0Ahttp://aisel.aisnet.org/cais%0Ahttp:

//aisel.aisnet.org/cais/vol39/iss1/7.

Farwick, M. et al., 2016. A situational method for semi-

automated Enterprise Architecture Documentation.

Software and Systems Modeling, 15(2).

Florez, H., Sánchez, M. & Villalobos, J., 2016. A catalog

of automated analysis methods for enterprise models.

SpringerPlus, 5(1). Available at:

https://www.scopus.com/inward/record.uri?eid=2-s2.0

-84962374165&doi=10.1186%2Fs40064-016-2032-9

&partnerID=40&md5=9cbe5683010939375dbd92417

b68553b.

Gartner Inc., 2017. Enterprise Architecture Definition.

https://www.gartner.com. Available at:

https://www.gartner.com/it-glossary/enterprise-

architecture-ea/.

Glazner, C.G., 2011. Enterprise transformation using a

simulation of enterprise architecture. Journal of

Enterprise Transformation, 1(3), pp.231–260.

Available at: https://www.scopus.com/inward/record

.uri?eid=2-s2.0-

84971449078&doi=10.1080%2F19488289.2011.6013

98&partnerID=40&md5=374ea2c633d670d78b138ff9

3408fbc6.

Group, T.O., 2009. The Open Group Architecture

Framework TOGAF

Version 9, Basharat Hussain.

Hauder, M., Matthes, F. & Roth, S., 2012. Challenges for

automated enterprise architecture documentation. In

Trends in Enterprise Architecture Research and

Practice-Driven Research on Enterprise

Transformation. Springer, pp. 21–39.

Huber, M. et al., 2017. The Influence of Shadow IT Systems

on Enterprise Architecture Management Concerns. In

European, Mediterranean, and Middle Eastern

Conference on Information Systems. Springer, pp. 461–

477.

IBM, 1978. Business System Planning: Information System

Planning Guide,

Jugel, D., Sandkuhl, K. & Zimmermann, A., 2016. Visual

analytics in Enterprise Architecture Management: a

systematic literature review. In International

Conference on Business Information Systems. Springer,

pp. 99–110.

Kim, T.-Y. et al., 2006. A modeling framework for agile

and interoperable virtual enterprises. Computers in

Industry, 57(3), pp.204–217. Available at:

https://search.proquest.com/docview/235768964?acco

untid=8144.

Korhonen, J.J. et al., 2016. Adaptive Enterprise

Architecture for the Future: Towards a

Reconceptualization of EA. Informatics (CBI), 2016

…, pp.272–281. Available at:

http://ieeexplore.ieee.org/abstract/document/7780323/

[Accessed April 11, 2017].

Kotusev, S., 2017. Conceptual Model of Enterprise

Architecture Management. International Journal of

Cooperative Information Systems.

Kotusev, S., 2016. The History of Enterprise Architecture:

An Evidence-Based Review. Journal of Enterprise

Architecture, 12(1), pp.29–37.

KPMG, 2016. Now or never - 2016 Global CEO Outlook,

Available at: https://home.kpmg.com/content/dam/

kpmg/pdf/2016/07/ch-2016-global-ceo-outlook.pdf.

Lange, M., Mendling, J. & Recker, J., 2016. An empirical

analysis of the factors and measures of Enterprise

Architecture Management success. European Journal

of Information Systems, 25(5), pp.411–431. Available

at: https://www.scopus.com/inward/record.uri?eid=2-

s2.0-

85016276209&doi=10.1057%2Fejis.2014.39&partner

ID=40&md5=a115806775e77eee39796a426d1ec930.

Lapalme, J. et al., 2016. Exploring the future of enterprise

architecture: A Zachman perspective. Computers in

Industry, 79, pp.103–113. Available at:

http://dx.doi.org/10.1016/j.compind.2015.06.010.

Lau, J.H. et al., 2011. Automatic labelling of topic models.

In Proceedings of the 49th Annual Meeting of the

Association for Computational Linguistics: Human

Language Technologies-Volume 1. Association for

Computational Linguistics, pp. 1536–1545.

Nikpay, F., Ahmad, R.B., et al., 2017. An effective

Enterprise Architecture Implementation Methodology.

Information Systems and e-Business Management,

pp.1–36. Available at: https://www.scopus.com

/inward/record.uri?eid=2-s2.0-85009446528&doi=10.

1007%2Fs10257-016-0336-5&partnerID=40&md5=

aa76b6715ccce7d27922eb974899a2e7.

Nikpay, F., Ahmad, R. & Yin Kia, C., 2017. A hybrid

method for evaluating enterprise architecture

implementation. Evaluation and Program Planning,

60, pp.1–16. Available at: https://www.scopus.com

/inward/record.uri?eid=2-s2.0-84988473835&doi=

10.1016%2Fj.evalprogplan.2016.09.001&partnerID=4

0&md5=39a635a5e1ff07077303805df3387f63.

Nugraha, D.C.A. et al., 2017. A business architecture

modeling methodology to support the integration of

primary health care: Implementation of primary health

care in Indonesia. Internetworking Indonesia Journal,

9(1), pp.39–45. Available at: https://www.scopus.com

/inward/record.uri?eid=2-s2.0-85017347432&

partnerID=40&md5=37629637e68808569da5e3eca12

b63c7.

Petersen, K., Vakkalanka, S. & Kuzniarz, L., 2015.

Guidelines for conducting systematic mapping studies

in software engineering: An update. Information and

Software Technology, 64, pp.1–18. Available at:

http://dx.doi.org/10.1016/j.infsof.2015.03.007.

Petticrew, M. & Roberts, H., 2008. Systematic reviews in

the social sciences: A practical guide, John Wiley &

Sons.

ICEIS 2018 - 20th International Conference on Enterprise Information Systems

646

Pinto, V.A., de Rezende Rohlfs, C.L. & Parreiras, F.S.,

2014. Applications of Ontologies in Enterprise

Modelling: A Systematic Mapping Study,

QualiWare, 2017. QualiWare Enterprise Architecture.

Center of Excellence. Available at:

https://coe.qualiware.com/ea-topic-model/.

Rahimi, F., Gøtze, J. & Møller, C., 2017. Enterprise

architecture management: Toward a taxonomy of

applications. Communications of the Association for

Information Systems, 40(1).

Ramage, D. & Rosen, E., 2011. Stanford topic modeling

toolbox.

Rouhani, B.D. et al., 2015. A systematic literature review

on Enterprise Architecture Implementation

Methodologies. Information and Software Technology,

62, p.1. Available at:

https://search.proquest.com/docview/1668188955?acc

ountid=8144.

Saint-Louis, P. & Lapalme, J., 2016. Investigation of the

Lack of Common Understanding in the Discipline of

Enterprise Architecture : A Systematic Mapping Study.

Proceedings - IEEE International Enterprise

Distributed Object Computing Workshop, EDOCW,

2016–Septe, pp.74–82.

Schneider, A.W., Schulz, C. & Matthes, F., 2013. Goals in

Enterprise Architecture Management--Findings from

Literature and Future Research Directions. In Business

informatics (cbi), 2013 ieee 15th conference on. IEEE,

pp. 284–291.

Simon, D., Fischbach, K. & Schoder, D., 2013. An

Exploration of Enterprise Architecture Research.

Communications of the Association for Information

Systems, 32(1), pp.1–71. Available at:

http://aisel.aisnet.org/cais/vol32/iss1/1.

Spewak, S.H. & Hill, S.C., 1993. Enterprise architecture

planning: developing a blueprint for data, applications

and technology, QED Information Sciences, Inc.

Turney, P.D. & Pantel, P., 2010. From Frequency to

Meaning: Vector Space Models of Semantics. Journal

of Artificial Intelligence Research, 37, pp.141–188.

Wißotzki, M. & Sonnenberger, A., 2012. Enterprise

architecture management-state of research analysis & a

comparison of selected approaches. actice of, p.37.

Zachman, J.A., 1987. A framework for information systems

architecture. IBM systems journal, 26(3), pp.276–292.

Understanding Enterprise Architecture with Topic Modeling

647

APPENDIX

Topic

(number &

label)

Coverage

of the

body of

literature

5 most common words (word instances per

topic)

Present in the

content

analysis of

Simon et at.

(2013)

Present in

topics of

Saint-Louis

& Lapalme

(2016)

2 Business

Process

0,60% process (56,63), basis (12,57), BusinessProcess

(8,37), financial (6,81), integrate (5,12)

Included in

the Business

Architecture

11 Framework 0,75% framework (117,89),

EnterpriseArchitectureFramework (8,12),

inte

rated

(

7,91

)

, structured

(

3,95

)

, ade

uate

(

2,99

)

No EA-

Framework

29 ERP Model 0,74% model (17,80), erp (12,69), develop (7,20),

companies (6,36), EnterpriseResourcePlanning

(6,17)

No No

52 Management 0,68% management (51,17), managers (14,90),

InformationSecurity (9,61), activities (8,79),

information (7,15)

Management

Related to

EA-Principles

102 Design 0,64% design (31,78), level (10,63), BusinessProcesses

(10,82), systems (9,75), component (9,71)

Similar to the

Documentatio

hase

EA-Design

118 Modelling

Methodology

0,62% modeling (58,62), HealthCare (17,55),

methodology (13,13), ModelingApproaches (4,85),

rapidly (4,60)

Modelling

element of the

Methodology

Related to

EA-Principles

and EA-

Strategy

211 Analysis

Method

0,61% method (55,63), analysis (26,92), CaseStudy (5,66),

estimates (3,87), technique (3,44)

Analysis

phase

Related to

EA-Principles

and EA-

Strategy

227 Meta

Model

0,78% metamodel (20,41), analysis (13,99), models (8,53),

perform (7,31), metamodels (6,81)

Meta Model

element of the

Methodology

319 Model 0,76% model (128,19), integrated (5,95), evolution (3,14),

contributes

(

2,96

)

, s

stems

(

2,28

)

Layers EA-Model

355 Service 0,69% service (43,66), services (39,65), resources (7,23),

user

(

6,36

)

, infrastructure

(

4,86

)

No No

Stopword list: enterprise, architecture, set, advances, number, presents, difficult, research, better, using, study, significant,

important, conducted, paper, findings, contribute, approach, common, general, source, right, based, attempts, diverse, large,

way, ensure, proposed, presented, web, proposes, allow, events, main, require, core, related, created, area, implemented,

emerged, numerous, use, private, defines, cover, scope, provided, recent, years, various, elements, includes, specific, ea, aim,

respect, different, multiple, small, nowadays, line, focuses, defined, major, new, increasing, insights, high, growing, helps,

poor, issues, gap, just, existing, result, simple, make, approaches, extend, results, base, studies, key, used, step, article, matter,

order, meet, enterprises, does, case, available, enable, business, review, future, provides, importance, means, needs, time,

today's, field, initiatives, network, needed, need, group, called, open, uses, best, lack, organization's, e.g, useful, enterprise's,

actual, making, work, setting, test, known, fact, typically, quickly, intended, shown, force, including, shows, primary, active,

describes, fully, bring, increase, allows, enables, era, takes, offer, offers, presenting, calls, range, example, today, real, works,

propose, consider, previous, cope, requires, focus, novel, suggested, introduced, establish, formal, highly, terms, makes, gives,

joint, early, gain, viewed, development, help, contribution, question, questions, contributions, utilised, utilized, second, broad,

emphasis, papers, suggests, concluded, described, concludes, plays, discusses, huge, continues, play, far, easy, given, applied,

follow, ongoing, ii, manner, so-called, issue, ad, hand, iii, i, later, taken, exist, like, versus, old, run, lot, starts, box, vast, end,

short, comes, going, left, author, authors, et, performance, lies, little, seen, ways, whilst, and/or, is/it

ICEIS 2018 - 20th International Conference on Enterprise Information Systems

648