Collecting and Integrating Unstructured Information into Enterprise
Architecture Management: A Systematic Literature Review
Robert Ehrensperger, Clemens Sauerwein and Ruth Breu
Institute of Computer Science, University of Innsbruck, Technikerstr. 21A, 6020 Innsbruck, Austria
Keywords: Unstructured Information, Information Collection, Enterprise Architecture Management.
Abstract: In the age of digital transformation, strategic IT alignment is becoming a primary driver for economic success.
In this context, the optimization of strategic IT alignment plays a key role in enterprise architecture
management (EAM). A successful EAM strategy depends on the quantity and quality of the available
information within the enterprise architecture (EA) models. EA information about the functional scope of
software solutions and its supported business processes is often available only in an unstructured form.
Automatic acquisition of this information assists companies in the design of target architectures. In recent
years, new technologies have been introduced that facilitate the use of unstructured information. The research
at hand discusses these new technologies and emerging challenges. Furthermore, it provides a systematic
literature review of the current state of research on collecting and integrating unstructured information into
EAM.
1 INTRODUCTION
Through the ongoing digitalization, the importance of
strategic IT alignment is increasing because it
supports the ability of companies to transform
themselves according to their business strategy
(Valenduc and Vendramin, 2017; Luftman and Brier,
1999). The principal objective of enterprise
architecture management (EAM) is to optimize this
strategic IT alignment (Farwick et al., 2016).
Therefore, EAM provides frameworks that allow
alignment of information systems and underlying IT
infrastructure with the business capabilities of an
enterprise (Babar and Yu, 2015).
The success of EAM crucially depends on the
quantity and quality of the available information
within the enterprise architecture (EA) models
(Fischer et al., 2007). These models represent EA
dimensions such as business, application, technology,
and information as well as the complex relationships
between them (Buschle et al., 2011; Lankhorst,
2009). Moreover, these models allow us to gain a
comprehensive understanding of an EA and are
essential to communicate the required changes to
stakeholders (Farwick et al., 2011). Enterprise
architecture management is designed to manage
necessary changes within companies’ IT landscapes.
It also allows us to improve the alignment of business
and IT and to increase the availability of IT systems
(Langenberg and Wegmann, 2004; Ross et al.,2006;
Ross, 2003).
The collection of additional unstructured
information from enterprise-external sources might
introduce new opportunities in EAM (Becker et al.,
2009; Becker et al., 2011). Enterprise-external
information sources include all sources that originate
outside of an internal enterprise environment and
reflect changes coming from the real world that are
relevant for enterprise architects. For example,
enhancing EAM with vendor product information,
such as license, product-life-cycle information, and
the functional scope of software solutions, might
allow improving the EA planning activities of
enterprise architects (Farwick et al., 2013). Also,
collecting announcements that appear in enterprise-
external information sources such as online blogs
about security vulnerabilities of IT systems might
allow for faster implementation of required EA
changes (Martin, 2008).
Technological improvements such as big data
(Bakshi, 2012) and artificial intelligence (O'Leary,
2013) have unlocked new potential for examining
unstructured information and drawing conclusions
from this information. However, research and
practice have not yet produced a comprehensive
728
Ehrensperger, R., Sauerwein, C. and Breu, R.
Collecting and Integrating Unstructured Information into Enterprise Architecture Management: A Systematic Literature Review.
DOI: 10.5220/0009339107280737
In Proceedings of the 22nd International Conference on Enterprise Information Systems (ICEIS 2020) - Volume 2, pages 728-737
ISBN: 978-989-758-423-7
Copyright
c
 2020 by SCITEPRESS – Science and Technology Publications, Lda. All rights reserved

understanding of how to collect unstructured
information for EAM based on these technologies.
Moreover, it is unclear to what extent these
technologies are used within EAM. This unclearness
motivated us to design the systematic literature
review at hand. The main research goal of this paper
is to provide a comprehensive overview of existing
research that deals with collecting and integrating
unstructured information into EAM and outlines
potential research challenges. Therefore, we
conducted a systematic literature review based on the
snowballing methodology (Wohlin, 2014). First, we
defined a start set of five papers, on which several
iterations of forward and backward snowballing were
performed. In doing so, we examined 384 papers.
This procedure resulted in a final set of 20 papers,
which we comprehensively analyzed.
The remainder of this paper is structured as
follows. Section 2 provides a summary of related
work. Section 3 outlines the applied research
methodology. Section 4 describes the results of the
snowballing procedure and the classification of
identified papers. Section 5 discusses the key findings
of the literature review concerning our research goal
and outlines potential research limitations. Finally,
Section 6 concludes our contribution and provides an
outlook on future work.
2 RELATED WORK
A few researchers have addressed the topic of
automated information collection for EAM. Farwick
et al. (Farwick et al., 2013) stated that information
sources such as network scanners and monitors,
configuration management databases, project
portfolio management tools, enterprise service buses,
change management tools and license management
tools can deliver valuable information for EAM. The
automated collection of information with
vulnerability scanners can also provide useful
information (Buschle et al., 2011). However,
according to multiple authors (Grunow et al., 2013;
Buschle et al., 2011; Farwick et al., 2011),
insufficient research has been conducted on the
analysis of potentially relevant information sources.
In summary, there exists related work on the
collection of structured information for EAM.
However, to the best of our knowledge, only a few
scientific investigations on the identification and
collection of relevant unstructured information for
EAM have been conducted. Additionally, the existing
research did not focus on relevant enterprise-external
information sources for EAM.
3 APPLIED RESEARCH
METHODOLOGY
This systematic literature review was finished in
December 2019. It is built on the snowballing
methodology (Wohlin, 2014), which is based on the
ideas of Webster and Watson (Webster and Watson,
2002). A comparison of the snowballing
methodology with the conventional database search
did not show any significant differences regarding the
results (Jalali and Wohlin, 2012). Moreover, applying
the snowballing methodology offers advantages in
obtaining information (Hendriks et al., 1992), such as
the ability to uncover hidden aspects (Atkinson and
Flint, 2001).
Based on the guidelines by Wohlin (Wohlin,
2014), our systematic literature study can be divided
into three steps. The first is the definition of start set.
The second is execution of snowballing iterations,
including both backward snowballing (i.e., looking at
the references of a paper under investigation), and
forward snowballing (i.e., identifying new papers
citing the paper being investigated). The last step of
our research methodology was analysis and
classification of the final set of papers.
3.1 Definition of Start Set
Investigations showed that Google Scholar finds
significantly more citations than other search engines
such as Web of Science and Scopus (Martín-Martín
et al., 2018). Accordingly, Google Scholar provides a
reliable data source to extract citation information.
Since the results of the snowballing methodology
crucially depend on citation information, we used
Google Scholar as search engine.
The first step of the snowballing methodology
consists of the definition of a start set. According to
Wohlin (Wohlin, 2014), there is no common
methodology for defining a start set. Therefore, we
searched for relevant publications focusing on
“enterprise architecture management” and
“unstructured information.” This initial search was
conducted systematically with the tool “Publish or
Perish” (Harzing, 2019), which allows users to create
complex and reproducible search queries by
accessing the Google Scholar API (application
programming interface). For our search, we specified
that the keyword “enterprise architecture
management” must appear in the title, and
“unstructured information” somewhere in the content
of the publication.
This procedure delivered a set of 11 papers. In
order to obtain only relevant publications, we first
Collecting and Integrating Unstructured Information into Enterprise Architecture Management: A Systematic Literature Review
729

analyzed the title, abstract, and keywords. If this
preliminary analysis identified the paper as
potentially relevant for our research goal, we
conducted a partial reading of the papers. During
reading of papers, we applied inclusion and exclusion
criteria to select papers for the start set. We included
papers that were accessible in full text, written in
English or German, and focused on the collection of
unstructured information for EAM. Any publication
that did not match any of the inclusion criteria was
excluded. We also excluded papers published before
2014, duplicates and publications dealing with big
data analytics not relating to EAM. The exclusion
criteria overrode the inclusion criteria; in other words,
if a publication met an exclusion criterion, it was
excluded even if it met one or more inclusion criteria.
After applying the inclusion and exclusion criteria,
we considered all remaining publications.
Ultimately, the first step of the applied research
methodology yielded a start set of five papers. These
papers are (Hacks and Saber, 2016; Rosina, 2015;
Roth, 2014; Schmidt et al., 2014b) and (Vanauer et
al., 2015).
3.2 Execution of Snowballing Iterations
In the second step of our systematic literature review,
we executed several iterations of snowballing, each
consisting of both forward and backward
snowballing, until no new papers were found. In total
four iterations of snowballing were executed, and 384
papers were examined. After each iteration, we again
analyzed the papers’ title, abstract, and keywords,
conducted a partial reading of papers deemed
potentially relevant, and applied the inclusion and
exclusion criteria (cf. section 3.1). After four
iterations no further relevant papers could be
identified, which terminated the snowballing
iterations. This procedure resulted in a final set of 20
papers.
3.3 Analysis and Classification of the
Final Set of Papers
We analyzed the remaining 20 publications according
to the following classification categories: (a)
bibliographic information, (b) applied research
methodology, (c) addressed research topics, and (d)
identified challenges. Each of the analyzed papers
was read at least by two authors of the publication at
hand.
The bibliographic information category includes
general information regarding the publication, such as
title, author, date, venue, and type of publication. The
research methodology category includes the
description and categorization of the applied research
methodologies (e.g., case study, survey, subjective/
argumentative study, descriptive/interpretative study,
experiment). This categorization system is derived
from (Galliers and Land, 1987). The addressed
research topic category classifies publications
according to the primary research topics they address.
We followed an iterative process, meaning that we
created a research topic category whenever at least
two publications were focusing on the same research
topic. Finally, we listed and sorted the research
challenges identified by the authors in order to
provide a comprehensive overview of addressed and
open research challenges. In doing so, all the authors
of the paper at hand discussed the mentioned
challenges and drew different categories out of it.
4 RESULTS
The applied snowballing methodology resulted in a
final set of 20 papers (Diefenthaler, 2016; Farwick et
al., 2016; Fittkau et al., 2015; Hacks et al., 2016;
Holm et al., 2014; Johnson et al.,2016; Kirschner and
Roth, 2014; Möhring and Schmidt, 2015; Ortmann et
al., 2014; Rosina, 2015; Roth, 2014; Roth and
Matthes, 2014; Schmidt et al., 2014a; Schmidt et al.,
2014b; Välja et al., 2016; Välja et al., 2015; Vanauer
et al., 2015; Zimmermann et al., 2016; Zimmermann
et al., 2017; Zimmermann et al., 2015). In this section,
the content of these papers and the results of the
classification are discussed.
4.1 Bibliographic Information
In the first step, the bibliographic information was
investigated. The number of publications remained
roughly constant between 2014 and 2016, averaging
more than six publications per year. A total of 19
publications between 2014 and 2016 were identified,
including three journal papers. Only one publication
fulfilling the criteria was found for 2017, and none for
2018 or 2019.
4.2 Research Methodology
In some publications, multiple research
methodologies are applied. The most theoretical work
in this field has used interpretive and argumentative
methods. Fourteen publications used a descriptive
research method, and nine of these focused on
describing frameworks for EAM. The second most
commonly used methodology was the subjective
ICEIS 2020 - 22nd International Conference on Enterprise Information Systems
730

approach, used in six publications. Three authors used
the research methodology experiment whereby they
tested approaches to automate EA modeling and its
visualization. Furthermore, two case studies were
identified evaluating the benefits of methods and
tools in an industrial setting. Moreover, surveys were
not used at all as a research methodology. However,
surveys provide a better description of the general
population (Rea and Parker, 2014) compared to the
other described research methodologies.
4.3 Research Topics
The identified papers address several research topics
regarding the collection of unstructured information
for EAM. Our investigations yielded a
comprehensive overview of the current state of
research in this field. Ultimately, it resulted in six
different categories, which are shown in Figure 1. In
this section, we will discuss each of these categories
in detail.
Figure 1: Main research topics found in the review.
4.3.1 Trigger Events
The first category includes trigger events, which are
required to start the information collection process.
For frequently changing information objects, these
triggers play a crucial role in the timeliness of the
collected information. Farwick et al. (Farwick et al.,
2016) and Roth (Roth, 2014) identified several
relevant trigger events. These events are classified
into three groups: (i) manual trigger events, (ii)
trigger events from enterprise-internal tools, and (iii)
trigger events from the EA repository. However,
existing research does not take into account
enterprise-external trigger events.
4.3.2 Information Types
The second category, which contained six papers,
addresses the different types of unstructured
information for EAM. The publications of Vanauer et
al., Rosina, Hacks, and Saber (Vanauer et al., 2015;
Hacks et al., 2016; Rosina, 2015) describe different
relevant information types such as documents,
spreadsheets, and presentations as well as general
applications and processes. Schmidt et al. (Schmidt et
al., 2014a) focused on these information types from
two different perspectives, the (i) static and (ii)
dynamic perspectives. These are shown in Figure 2.
The static perspective operates mainly on structured
information and supports strategic decision making.
The dynamic perspective focuses on highly volatile
semi- and unstructured information such as log files.
Leveraging both perspectives would allow us to run a
descriptive, prescriptive, and predictive analysis for
EAM. These analysis opportunities support EAM not
only in long-term strategic decisions but also in
imminent tactical or even operational decisions.
Figure 2: A static and dynamic perspective on EAM
information (Schmidt et al., 2014a).
Moreover, the research discusses linked information.
For example, the authors (Zimmermann et al., 2017;
Ortmann et al., 2014) highlight this type of
information. Linking different information such as
servers, applications, interfaces, and their supported
business processes enables enterprises to reach value-
added conclusions, such as the identification of
redundant IT systems in the EA.
4.3.3 Information Quality
The third category addresses the quality of the
collected information. Hacks and Saber (Hacks and
Saber, 2016) argued that better quality can be
achieved through preprocessing and preparation of
the information. Ortmann et al. (Ortmann et al., 2014)
raised the possibility of increasing information
quality by conducting analysis directly on the original
enterprise-external information sources in their
diverse formats and terminologies. The advantage of
this approach is that the information can be made
available in a timely fashion and thereby is of higher
quality. An automated information collection process
is less error-prone than a manual process. Thus, the
automation of the information collection process
leads to better information quality, according to Holm
et al. (Holm et al., 2014).
Collecting and Integrating Unstructured Information into Enterprise Architecture Management: A Systematic Literature Review
731

4.3.4 Information Sources
The fourth category concerns relevant information
sources for EAM. In total, seven publications were
identified that discuss these information sources,
which can be divided into enterprise-internal and
enterprise-external sources. First, many researchers
such as Farwick et al. (Farwick et al., 2016), Johnson
et al. (Johnson et al.,2016), Välja et al. (Välja et al.,
2016), Diefenthaler (Diefenthaler, 2016), Schmidt et
al. (Schmidt et al., 2014a), and Fittkau et al. (Fittkau
et al., 2015) identified and described different
enterprise-internal information sources. These
sources can be categorized according to the
components of IT systems: hardware (e.g., systems,
infrastructure), software (e.g., runtime information,
vulnerability), databases (e.g., wiki’s, change-,
license-, and portfolio-management tools), networks
(e.g., network monitors and scanners), and
procedures (e.g., process state information, enterprise
service bus). This structure is derived from Rainer
and Cegielski (Rainer and Cegielski, 2013).
Six of the seven publications focused on
enterprise-internal information sources. Only one
publication discussed enterprise-external information
sources; Zimmermann et al. (Zimmermann et al.,
2017) underlined that in addition to enterprise-
internal sources, enterprise-external sources might
also improve EAM. However, the authors did not
mention any specific information sources.
4.3.5 EA Models
The fifth category discusses EA models and the use
of unstructured information sources to create or
maintain existing models. These models allow us to
visualize EA information, which supports us in
understanding and working with EAs (Fischer et al.,
2007). In our analysis, we distinguish between the
following two categories: (i) creation of EA models,
and (ii) maintenance of EA models. In total, we
identified six publications that addressed EA models.
First, in order to automate the creation of EA
models, Johnson et al. (Johnson et al.,2016) proposed
the use of machine learning techniques. They
considered the use of dynamic Bayesian networks
because such networks can capture fuzzy information
that describes EAs. Välja et al. (Välja et al., 2016)
automatized the creation of EA models from multiple
heterogeneous information sources. To do so, they
used the Joint Directors of Laboratories (JDL)
framework (Välja et al., 2016; Liggins, 2008;
Steinberg and Bowman, 2017), which facilitates the
fusion of information on different granular levels.
Second, for the maintenance of EA models,
Kirschner and Roth (Kirschner and Roth, 2014)
merged different EA models for a single EA
repository. In order to reach this goal, they described
merge algorithms, which detect EA model conflicts
and generate resolution tasks. Roth and Matthes
(Roth and Matthes, 2014) presented a concept that
allows the differences between EA models to be
analyzed and visualized. Fittkau et al. (Fittkau et al.,
2015) introduced an approach that utilizes
information system monitoring to improve
consistency between EA models and real information
systems. Furthermore, (Zimmermann et al., 2016)
detailed a process that makes it possible for
information sources to be integrated continuously
with an EA model.
4.3.6 Supporting Technologies
Finally, a remarkable number of publications discuss
technologies that support the collection of
unstructured information for EAM. This topic will
attract more attention in the future because many new
technologies are on the rise (Bakshi, 2012; O'Leary,
2013). Generally speaking, to make it more
comprehensible, these technologies can be
categorized according to their nature into (i) big data
technologies, (ii) semantic technologies, and (iii)
service-oriented technologies. Eight publications
were included in the following overview.
First, big data technologies allow us to collect
large quantities of EAM-relevant information with
varying structure and to conduct near-real-time
analysis on, for example, architectural information
contained in many infrastructure components
(Zimmermann et al., 2017; Schmidt et al., 2014a;
Provost and Fawcett, 2013). Thus, a considerable
number of researchers have focused on the supportive
role of big data in the collection of unstructured
information for EAM (Möhring and Schmidt, 2015;
Hacks and Saber, 2016). Hacks and Saber (Hacks and
Saber, 2016) explain the state of the art of the usage
of big data technologies for EAM. They evaluated
how different big data frameworks (e.g., Hadoop
(Ghazi and Gangodkar, 2015)) can be used to support
specific EAM requirements. A methodology to
deploy big data for information collection was
introduced by Vanauer et al. (Vanauer et al., 2015).
This publication provides guidelines for
implementing big data technologies in the EAM
context.
Second, the use of semantic technologies for
information collection in EAM was proposed by
Ortmann et al. (Ortmann et al., 2014) and Rosina
ICEIS 2020 - 22nd International Conference on Enterprise Information Systems
732

(Rosina, 2015). The idea behind semantic
technologies is to leverage the semantic value of the
collected information. The responsibility to keep the
original information up to date remains with the
information’s owner. This concept allows accessing
current information from diverse information sources
independently of the original format and terminology.
Rosina showed how, based on the use of these
semantic technologies, EA information could be
collected, formalized, and integrated into EAM.
Third, in order to enable the collection and
integration of a growing diversity of information for
EAM, a concept that uses service-oriented
technologies was presented by Zimmermann et al.
(Zimmermann et al., 2015). The goal of this concept
is to foster digital transformation based on a holistic
EAM approach. This approach integrates information
from the Internet of Things into EAM. Zimmermann
et al. focus on the extension of a static enterprise-
internal architecture to accommodate the flexible and
adaptive digitization of new information sources that
come from an enterprise-external environment
(Zimmermann et al., 2016; Zimmermann et al.,
2017). This extension is realized with the help of
microservice technologies that enable the integration
of information sources into an EA.
4.4 Challenges
The authors of the publications examined in this
literature review identified a variety of challenges. In
order to create a systematic overview on these
challenges, we defined the following categories: (i)
exploitation of information, (ii) information quality,
and (iii) EA governance. They are briefly discussed
in the following section.
4.4.1 Exploitation of Information
The work of Hacks and Saber (Hacks and Saber,
2016; Katal et al., 2013) identified challenges
regarding the exploitation of information for EAM,
including the difficulty of identifying relevant,
accurate information to support EAM decision
making. For example, enterprise architects have to
make decisions influencing the software release
management process of an EA. Therefore, it is
essential to have accurate information about the
restrictions of the different versions of software
applications to be released.
Besides identifying information, it was also
highlighted that unstructured information cannot be
uniformly analyzed and are more challenging to
process for big data technologies. More research is
required to investigate how big data technologies may
be used in the analysis of EAM-relevant unstructured
information. As objects of further investigation,
information about the functional scope of software
solutions and its supported business processes may be
used.
Moreover, Zimmermann et al. (Zimmermann et
al., 2015) noted that the integration of a vast number
of dynamically growing systems and services, such as
Microservices and the Internet of Things, presents a
considerable challenge for the scalability, extension,
and evolution of EA models. In this context, the
processing of raw information objects, such as the
output of network scanners, to fulfill their specific
purposes within EAM (e.g., into an EA model)
remains a challenge (Holm et al., 2014). One obstacle
is the difficulty of defining clear rules for this
processing. This was also emphasized by Välja et al.
(Välja et al., 2015) and Farwick et al. (Farwick et al.,
2016).
Furthermore, Ortmann et al. (Ortmann et al.,
2014) highlighted the challenge of identifying
relationships between different information objects in
order to derive new knowledge regarding the EA. For
example, business partners might be modeled as
information objects. For enterprise architects, it is
vital to know at what date the object business partner
is stored by which software application within the
EA. Therefore, the identification of relationships
between objects is essential.
4.4.2 Information Quality
Information quality plays a crucial role when
integrating unstructured information into EAM.
Unstructured information has to be reliable to be able
to draw the right conclusions. Holm et al. (Holm et
al., 2014) used network scanners to identify system
software, applications, and interfaces and assessed
the quality of the resulting information. They
discussed the challenges of tracking the quality of this
information over time in order to create a coherent
view.
Furthermore, Fittkau et al. (Fittkau et al., 2015)
and Roth (Roth, 2014) outlined the challenge of
maintaining consistency between EA models and
information sources. This problem might be traced
back to manual information collection processes,
which are still in place in many companies.
Finally, Farwick et al. (Farwick et al., 2016)
characterized the research into the information
quality issue as only a small island within the
literature, indicating a profound lack of research in
this field.
Collecting and Integrating Unstructured Information into Enterprise Architecture Management: A Systematic Literature Review
733

4.4.3 EA Governance
In regard to managing the changes required for the
integration of unstructured information into an EA,
several challenges on the EA governance level can be
identified. The involvement of stakeholders with
differing or even contradicting but interrelated areas-
of-interest was mentioned as a challenge by Rosina
(Rosina, 2015) and Schmidt et al. (Schmidt et al.,
2014b). For example, the integration of publicly
available unstructured information about customers
such as product feedback, opinions, or interests can
reveal the differing goals of stakeholders within a
company. A sales department might be interested in
improving its customer care activities, while the IT
department might want to reduce its operational
support resources.
Moreover, various stakeholders often use
different vocabulary to describe the same information
object. Thus, it is difficult to share, exchange, or
consolidate information about integration approaches
for EAM. Further research should focus on the topic
of how to facilitate the communication of enterprise
architecture transformations and the involvement of
different stakeholders.
5 DISCUSSION
In the following section, we discuss the key findings
and limitations of the research at hand.
5.1 Key Findings
Based on the results of our systematic literature
review and the subsequent analysis, we derived the
following five key findings.
Key Finding 1: There is a profound lack of surveys
and case studies regarding requirements and sources
for collecting unstructured information for EAM.
Our systematic literature review and the related
work indicated a lack of surveys and case studies
investigating the collection for unstructured
information for EAM (cf. section 4.2). This lack
suggests that researchers are not leveraging the
advantages of surveys and case studies in this field.
Case studies hold the potential to gain insights into
many details that would not usually be easily obtained
by other research methodologies. The results of case
studies are usually richer and of greater depth than
can be obtained through other experimental designs.
For example, a case study might clarify the potential
use cases that arise from collecting unstructured
information for EAM. The advantage of a survey is
the ability to gather qualitative feedback about the
need for collecting and integrating unstructured
information into EAM. Surveys may help to assess
the expected utility value of collecting unstructured
information for EAM. In summary, more research,
such as a survey and a case study, is needed to
determine the requirements for the implementation of
tools and frameworks for automated collection of
relevant unstructured information.
Key Finding 2: Leveraging dynamically changing
unstructured information holds the potential to
predict required EA changes in the future.
Our investigations showed that existing research
does not leverage dynamically changing unstructured
information (cf. section 4.3.2). To identify the
required EA adjustments, tracing of dynamically
changing information is helpful. For example, rapid
growth in the size of log file entries might lead to a
lack of storage resources. The footprint of how users
interact with a GUI (graphical user interface) can
point out the popularity of different functionalities.
This dynamic un- and semistructured information
might pave the way to predict required EA changes in
the future. Moreover, leveraging this information
allows us to react more quickly to changing EA
vulnerabilities, thus giving us more time to make the
needed adjustments to the EA. Linking different
information types (e.g., correlating highly volatile
information with static information) would also
enable identification of legacy IT systems. This
identification can be made by linking real-time
operational information and static EA model
information of a particular IT system and outlining
potential mismatches between them.
Key Finding 3: Investigations regarding enterprise-
external information sources or trigger events for
EAM and their relationships are missing.
This review confirmed that the majority of EAM-
relevant information originates from enterprise-
internal sources (cf. section 4.3.4). The authors of
existing research did not mention any enterprise-
external information sources that are used within
EAM. However, the need to collect enterprise-
external sources is explicitly expressed in literature
(Zimmermann et al., 2017). This fact reveals the need
for more research focusing on enterprise-external
information sources for EAM. Our investigations
showed that currently, information collection
processes are triggered mainly by enterprise-internal
events. However, for the collection of enterprise-
external information sources, it can also be necessary
ICEIS 2020 - 22nd International Conference on Enterprise Information Systems
734

to leverage trigger events from an enterprise-external
environment. The question of which trigger events for
enterprise-external information sources are adequate
to trigger information collection processes
automatically remains open.
Key Finding 4: Technology is on the rise that tackles
long-standing challenges regarding the collection of
unstructured information for EAM.
In summary, this review discusses three general
technologies that enable the collection of
unstructured information for EAM (cf. section 4.3.6).
First, big data technologies can store and analyze
massive amounts of unstructured information. This
might allow us to handle the growing amount of
unstructured information from an enterprise-external
environment. Second, semantic technologies can
integrate highly dynamic information from diverse
information sources by keeping their timeliness. The
advantage of using these semantic technologies is that
the responsibility to maintain the source information
remains with the information’s owner, and EAM
could collect the information without needing to
spend effort on managing mass storages and
maintaining information quality. Moreover, service-
oriented technologies seem to be adequate to manage
a growing diversity of EAM information. It is
possible to easily reuse and adjust existing services to
change and extend the EA repository.
Key Finding 5: Information quality remains a crucial
issue for the collection of unstructured information
for EAM.
Only a few researchers highlighted and
considered the quality of the collected information
(cf. section 4.3.3). It is challenging to derive specific
further actions without adequate information quality.
However, there has not yet been sufficient research
into the required quality of unstructured information
for EAM. Accordingly, further investigations are
necessary to develop appropriate measures to
guarantee the necessary information quality for EAM.
5.2 Limitations
The research at hand might be limited by a (i)
selection bias of papers, (ii) false classification and
analysis, (iii) missing papers and (iv) limited
generalizability of results. In order to overcome (i),
this review is based on the well-established guidelines
by Wohlin et al. (Wohlin, 2014). A detailed
description of the implementation of these guidelines
can be found in Section 3. In order to overcome (ii),
we provided definitions of the classification criteria
and analyzed the papers based on them. Moreover,
each of the observed papers was read at least by two
authors of the publication at hand. Moreover, there is
the possibility that we (ii) missed relevant papers.
Since we applied the snowballing methodology, and
recent studies have shown that it delivers comparable
results to other research methodologies for
conducting systematic literature studies (Jalali and
Wohlin, 2012; Badampudi et al., 2015), the risk of
(iii) is at an acceptable level. Finally, there might be
the risk of (iv). We reduced this risk by applying the
snowballing methodology until no new papers were
found. In doing so, we provided a comprehensive
overview of existing research in the field.
6 CONCLUSION AND FUTURE
WORK
This paper presents a systematic literature review that
provides a comprehensive overview of existing
research into the collection and use of unstructured
information for EAM. It uses the snowballing
methodology to identify relevant literature and
classifies it according to the following criteria:
bibliographic information, research methodology,
research topics, and research challenges. In total, we
identified and classified 20 relevant publications. Our
investigations showed that there is a profound lack of
research regarding the requirements and sources for
collecting unstructured information for EAM. We
also determined that leveraging the dynamic
perspective of unstructured information enables the
prediction of required EA changes in the future.
However, we found that there has been little research
regarding enterprise-external information sources or
trigger events for EAM and their relationships. Our
investigations also showed that there is technology on
the rise that can help tackle long-standing challenges
regarding the collection of unstructured information.
Finally, we determined that there is a lack of research
regarding the required information quality of
unstructured information for use in EAM. In
summary, the amount of unstructured information is
continuously growing, and some early steps have
been undertaken to leverage this information for
EAM. However, the results of this review suggest that
enterprise-external information sources have not yet
been investigated in detail. Further research is
required to identify relevant enterprise-external
information that supports EAM. Because researchers
have not undertaken any surveys in this field, we plan
to conduct an exploratory survey. The goal of this
Collecting and Integrating Unstructured Information into Enterprise Architecture Management: A Systematic Literature Review
735

survey is to provide a comprehensive view of the
relevant enterprise-external information sources from
a research and practice perspective.
REFERENCES
Atkinson, R., & Flint, J. (2001). Accessing hidden and
hard-to-reach populations: Snowball research
strategies. Social research update, 33(1), 1-4.
Babar, Z., & Yu, E. (2015, June). Enterprise architecture in
the age of digital transformation. In International
Conference on Advanced Information Systems
Engineering (pp. 438-443). Springer, Cham.
Badampudi, D., Wohlin, C., & Petersen, K. (2015, April).
Experiences from using snowballing and database
searches in systematic literature studies. In Proceedings
of the 19th International Conference on Evaluation and
Assessment in Software Engineering (p. 17). ACM.
Bakshi, K. (2012, March). Considerations for big data:
Architecture and approach. In 2012 IEEE Aerospace
Conference (pp. 1-7). IEEE.
Becker, H., Naaman, M., & Gravano, L. (2009, June).
Event Identification in Social Media. In WebDB.
Becker, H., Naaman, M., & Gravano, L. (2011, July).
Beyond trending topics: Real-world event identification
on twitter. In Fifth international AAAI conference on
weblogs and social media.
Budgen, D., & Brereton, P. (2006, May). Performing
systematic literature reviews in software engineering.
In Proceedings of the 28th international conference on
Software engineering (pp. 1051-1052). ACM.
Buschle, M., Holm, H., Sommestad, T., Ekstedt, M., &
Shahzad, K. (2011, June). A Tool for automatic
Enterprise Architecture modeling. In International
Conference on Advanced Information Systems
Engineering (pp. 1-15). Springer, Berlin, Heidelberg.
Diefenthaler, P. I. (2016). Interactive Transformation Path
Generation in Enterprise Architecture Planning.
Farwick, M., Agreiter, B., Breu, R., Ryll, S., Voges, K., &
Hanschke, I. (2011). Requirements for automated
enterprise architecture model maintenance. In 13th
International Conference on Enterprise Information
Systems (ICEIS), Beijing.
Farwick, M., Breu, R., Hauder, M., Roth, S., & Matthes, F.
(2013, January). Enterprise architecture
documentation: Empirical analysis of information
sources for automation. In 2013 46th Hawaii
International Conference on System Sciences (pp.
3868-3877). IEEE.
Farwick, M., Schweda, C. M., Breu, R., & Hanschke, I.
(2016). A situational method for semi-automated
Enterprise Architecture Documentation. Software &
Systems Modeling, 15(2), 397-426.
Fischer, R., Aier, S., & Winter, R. (2007). A federated
approach to enterprise architecture model
maintenance. Enterprise Modelling and Information
Systems Architectures (EMISAJ), 2(2), 14-22.
Fittkau, F., Roth, S., & Hasselbring, W. (2015, May).
ExplorViz: Visual runtime behavior analysis of
enterprise application landscapes. AIS.
Galliers, R. D., & Land, F. F. (1987). Choosing appropriate
information systems research methodologies.
Communications of the ACM, 30(11), 901-902.
Ghazi, M. R., & Gangodkar, D. (2015). Hadoop,
MapReduce and HDFS: a developers perspective.
Procedia Computer Science, 48, 45-50.
Grunow, S., Matthes, F., & Roth, S. (2013). Towards
automated enterprise architecture documentation: Data
quality aspects of SAP PI. In Advances in Databases
and Information Systems (pp. 103-113). Springer,
Berlin, Heidelberg.
Hacks, S., & Saber, M. (2016). Supportive Role of Big Data
Technologies in Enterprise Architecture Management.
Full-scale Software Engineering/Current Trends in
Release Engineering, 13, 37.
Hendriks, V. M., Blanken, P., Adriaans, N. F. P., &
Hartnoll, R. (1992). Snowball sampling: A pilot study
on cocaine use. IVO, Instituut voor
Verslavingsonderzoek, Erasmus Universiteit
Rotterdam.
Holm, H., Buschle, M., Lagerström, R., & Ekstedt, M.
(2014). Automatic data collection for enterprise
architecture models. Software & Systems Modeling,
13(2), 825-841.
Harzing, A.-W. (2019). Publish or Perish (Version
6.34.6288.6798): Tarma Software Research. Available
under https://harzing.com/resources/publish-or-perish
[29.10.2019].
Jalali, S., & Wohlin, C. (2012, September). Systematic
literature studies: database searches vs. backward
snowballing. In Proceedings of the 2012 ACM-IEEE
international symposium on empirical software
engineering and measurement (pp. 29-38). IEEE.
Johnson, P., Ekstedt, M., & Lagerstrom, R. (2016,
September). Automatic probabilistic enterprise IT
architecture modeling: a dynamic bayesian network
approach. In 2016 IEEE 20th International Enterprise
Distributed Object Computing Workshop (EDOCW)
(pp. 1-8). IEEE.
Katal, A., Wazid, M., & Goudar, R. H. (2013, August). Big
data: issues, challenges, tools and good practices.
In 2013 Sixth international conference on
contemporary computing (IC3) (pp. 404-409). IEEE.
Kirschner, B., & Roth, S. (2014). Federated enterprise
architecture model management: collaborative model
merging for repositories with loosely coupled schema
and data. In Multikonferenz Wirtschaftsinformatik.
Langenberg, K., & Wegmann, A. (2004). Enterprise
architecture: What aspects is current research
targeting (No. REP_WORK).
Lankhorst, M. (2009). Enterprise architecture at
work (Vol. 352). Berlin: Springer.
Liggins, M. E. (2008). Handbook of Multi-Sensor Data
Fusion, ME Liggins, DL Hall, and J. Llinas, Eds.
Google LLC. Google scholar - bibliographic database.
https://scholar.google.de/, 09 2018. Accessed: 30-Oct-
2019.
ICEIS 2020 - 22nd International Conference on Enterprise Information Systems
736

Luftman, J., & Brier, T. (1999). Achieving and sustaining
business-IT alignment. California management
review, 42(1), 109-122.
Martín-Martín, A., Orduna-Malea, E., Thelwall, M., &
López-Cózar, E. D. (2018). Google Scholar, Web of
Science, and Scopus: A systematic comparison of
citations in 252 subject categories. Journal of
Informetrics, 12(4), 1160-1177.
Martin, R. A. (2008, November). Making security
measurable and manageable. In MILCOM 2008-2008
IEEE Military Communications Conference (pp. 1-9).
IEEE.
Möhring, M., & Schmidt, R. (2015). Daten-getriebene
Unternehmensarchitekturen im E-Commerce für das
präventive Retourenmanagement. INFORMATIK 2015.
O'Leary, D. E. (2013). Artificial intelligence and big
data. IEEE Intelligent Systems, 28(2), 96-99.
Ortmann, J., Diefenthaler, P., Lautenbacher, F., Hess, C., &
Chen, W. (2014). Unternehmensarchitekturen mit
semantischen technologien. HMD Praxis der
Wirtschaftsinformatik, 51(5), 616-626.
Provost, F., & Fawcett, T. (2013). Data Science for
Business: What you need to know about data mining
and data-analytic thinking. " O'Reilly Media, Inc.".
Rainer, R. K., Cegielski, C. G., Splettstoesser-Hogeterp, I.,
& Sanchez-Rodriguez, C. (2013). Introduction to
information systems: Supporting and transforming
business. John Wiley & Sons.
Rea, L. M., & Parker, R. A. (2014). Designing and
conducting survey research: A comprehensive guide.
John Wiley & Sons.
Rosina, P. C. (2015). Semantic Web Technologies for
Method Management: Model, Methodology and
Enterprise Architecture Integration.
Ross, J. W. (2003). Creating a strategic IT architecture
competency: Learning in stages.
Ross, J. W., Weill, P., & Robertson, D. (2006). Enterprise
architecture as strategy: Creating a foundation for
business execution. Harvard Business Press.
Roth, S. (2014). Federated Enterprise Architecture Model
Management (Doctoral dissertation, Technische
Universität München).
Roth, S., & Matthes, F. (2014). Visualizing differences of
enterprise architecture models. In International
Workshop on Comparison and Versioning of Software
Models (CVSM) at Software Engineering (SE). Kiel,
Germany.
Schmidt, R., Wißotzki, M., Jugel, D., Möhring, M.,
Sandkuhl, K., & Zimmermann, A. (2014a, September).
Towards a framework for enterprise architecture
analytics. In 2014 IEEE 18th International Enterprise
Distributed Object Computing Conference Workshops
and Demonstrations (pp. 266-275). IEEE.
Schmidt, R., Zimmermann, A., Möhring, M., Jugel, D., Bär,
F., & Schweda, C. M. (2014b, September). Social-
software-based support for enterprise architecture
management processes. In International Conference on
Business Process Management (pp. 452-462).
Springer, Cham.
Steinberg, A. N., & Bowman, C. L. (2017). Revisions to the
JDL data fusion model. In Handbook of multisensor
data fusion (pp. 65-88). CRC Press.
Valenduc, G., & Vendramin, P. (2017). Digitalisation,
between disruption and evolution. Transfer: European
Review of Labour and Research, 23(2), 121-134.
Välja, M., Korman, M., Lagerström, R., Franke, U., &
Ekstedt, M. (2016, September). Automated architecture
modeling for enterprise technology manageme using
principles from data fusion: A security analysis case.
In 2016 Portland international conference on
management of engineering and technology (PICMET)
(pp. 14-22). IEEE.
Välja, M., Lagerström, R., Ekstedt, M., & Korman, M.
(2015, September). A requirements based approach for
automating enterprise it architecture modeling using
multiple data sources. In 2015 IEEE 19th International
Enterprise Distributed Object Computing Workshop
(pp. 79-87). IEEE.
Vanauer, M., Böhle, C., & Hellingrath, B. (2015, January).
Guiding the introduction of big data in organizations: A
methodology with business-and data-driven ideation
and enterprise architecture management-based
implementation. In 2015 48th Hawaii International
Conference on System Sciences (pp. 908-917). IEEE.
Webster, J., & Watson, R. T. (2002). Analyzing the past to
prepare for the future: Writing a literature review. MIS
quarterly, xiii-xxiii.
Wieringa, R., Maiden, N., Mead, N., & Rolland, C. (2006).
Requirements engineering paper classification and
evaluation criteria: a proposal and a discussion.
Requirements engineering, 11(1), 102-107.
Wißotzki, M., & Sonnenberger, A. (2012). Enterprise
architecture management-state of research analysis & a
comparison of selected approaches. In 5th IFIP WG 8.1
Working Conference, PoEM 2012 (pp. 37-48).
Wohlin, C. (2014, May). Guidelines for snowballing in
systematic literature studies and a replication in
software engineering. In Proceedings of the 18th
international conference on evaluation and assessment
in software engineering (p. 38). ACM.
Zimmermann, A., Bogner, J., Schmidt, R., Jugel, D.,
Schweda, C., & Möhring, M. (2016). Digital enterprise
architecture with micro-granular systems and services.
Zimmermann, A., Schmidt, R., Sandkuhl, K., Jugel, D.,
Bogner, J., & Möhring, M. (2017, October). Decision
management for micro-granular digital architecture.
In 2017 IEEE 21st International Enterprise Distributed
Object Computing Workshop (EDOCW) (pp. 29-38).
IEEE.
Zimmermann, A., Schmidt, R., Sandkuhl, K., Wißotzki, M.,
Jugel, D., & Möhring, M. (2015, September). Digital
enterprise architecture-transformation for the internet
of things. In 2015 IEEE 19th International Enterprise
Distributed Object Computing Workshop (pp. 130-
138). IEEE.
Collecting and Integrating Unstructured Information into Enterprise Architecture Management: A Systematic Literature Review
737