Emerging Concepts and Trends in Collaborative Modeling: A Survey

Matthew Stephan

Dept. of Computer Science and Software Engineering, Miami University, 510 East High Street, Oxford Ohio, U.S.A.

Keywords:

Model-Driven Engineering, Modeling for the Cloud, Collaborative Modeling, Collaborative Software

Engineering.

Abstract:

Just as in other engineering disciplines, software engineering is well suited to collaboration; Having different

perspectives and diverse experiences strengthens engineering projects. Software modeling is a fundamental

aspect of software engineering and is becoming increasingly collaborative. Collaborative modeling approaches

are maturing and related research is growing signiﬁcantly. While surveys exist on collaborative modeling

tools and research, they are aimed at academics and can be verbose. In this article, we conduct a research

survey intended to provide practitioners and researchers an accessible and abstract at-a-glance perspective of

emerging trends and directions in collaborative modeling. We complete a systematic literature review, which

we crosscheck with existing surveys. To explicate trends in the last ﬁve complete years and overall trends,

we perform concept extraction and domain analysis by analyzing abstracts. We visualize these trends in word

clouds and trend charts, and provide insights. We hope this article helps spread awareness of collaborative

modeling trends and future directions, and educates practitioners and researchers.

1 INTRODUCTION

Software modeling is an integral part of software en-

gineering. For complex software systems, Model-

Driven Engineering (MDE) is one approach that helps

deal with accidental complexity and allows designers

to work with artifacts at an abstraction level closer to

their problem domains (Mellor et al., 2003). MDE

involves creating software models that act as the pri-

mary artifacts throughout all phases within the soft-

ware engineering life cycle. Whether organizations

are employing software models in formal MDE ap-

proaches or informally, such as in design decisions

and inception prototyping, software modeling is in-

creasingly prevalent in industry (Hutchinson et al.,

2011).

Engineering, by its nature, is suited to collabora-

tion as multiple minds with different perspectives can

serve only to improve results. This holds true for col-

laborative modeling, where many software engineers

work together to develop models. However, given

the advent of Semantic Web technologies (McIlraith

et al., 2001) and the increasing importance being

placed on collaboration in general, new collaborative

research projects are emerging. Thus, our goal is to

provide an abstract overview of recent collaborative

modeling research to identify temporal trends. While

other surveys exist on collaborative modeling, ours

provides a temporal perspective, discovers emerging

concepts, and identiﬁes trends. We present emerging

concepts in collaborative modeling research from 1)

the most recent complete 5 years (2012-2017) and 2)

all time, and contrast these to explicate trends and di-

rections. We additionally pick 2012 as a starting point

because it was the ﬁrst year of a dedicated workshop

on the subject (Ober et al., 2012), indicating a new fo-

cus and importance placed on collaborative modeling

by the research community.

We aim to answer the following questions we be-

lieve of use to practitioners and researchers,

1. What are the emerging concepts in collaborative

modeling within the last 5 complete years and all

time?

2. In what way has the focus on these concepts

changed over time?

To provide a foundation and the data for answer-

ing these questions we perform a systematic litera-

ture review (SLR) of collaborative modeling. This in-

cludes following a speciﬁc, constrained, and repeat-

able search process. To ascertain trends, we analyze

abstracts of papers to determine concept frequency

and conduct domain analysis (Paraschiv et al., 2015).

We visualize this analysis in the form of word clouds

240

Stephan, M.

Emerging Concepts and Trends in Collaborative Modeling: A Survey.

DOI: 10.5220/0007255502400247

In Proceedings of the 7th International Conference on Model-Driven Engineering and Software Development (MODELSWARD 2019), pages 240-247

ISBN: 978-989-758-358-2

and trend charts to illustrate temporal content evolu-

tion (Cui et al., 2010). We then provide an overview

of notable emerging trends and discuss these ﬁndings.

We begin with a presentation of related surveys in

Section 2, and follow with our survey protocol in Sec-

tion 3. We summarize our data in Section 4 includ-

ing prevalent topical words, quantitative data, and our

word clouds. In Section 5, we discuss the changing

trends and interpret the data to provide some of our

insights and thoughts. We conclude the paper in Sec-

tion 6.

2 RELATED WORK

In this section, we discuss related secondary stud-

ies/surveys to this paper. While there are several stud-

ies that provide different perspectives on collaborative

modeling (Renger et al., 2008; Franzago et al., 2017;

Rocha et al., 2011; Portillo-Rodr

ıguez et al., 2012),

none of them consider trends nor temporal aspects.

The most complete, recent, and relevant to ours is that

of Micro Franzago et al. (Franzago et al., 2017) pub-

lished in 2017. They classify existing techniques and

discuss 48 primary studies spanning 19 years. Our

survey has a different purpose and provides a different

perspective, that being a temporal focus and compar-

ison to provide an abstract at-a-glance view and dis-

cussion of trends. We were thus more inclusive than

they were, as we will describe in our protocol. How-

ever, we did cross-reference their paper to ensure we

did not miss any primary studies. As a result, our

study includes their 48 primary studies and more.

3 SURVEY PROTOCOL

While our goal was to provide an abstract act-a-glance

view of trends, we still attempted to follow estab-

lished guidelines for conducting software engineering

systematic literature reviews (Kitchenham and Char-

ters, 2007). This helps ensure that others can repeat

and reproduce our work.

3.1 Research Question

To help frame our research question, we use Petticrew

and Roberts’ PICOC criteria: (Petticrew and Roberts,

2008):

• Population:

1) Practitioners interested in comparing and em-

ploying collaborative modeling who want a tem-

poral perspective illustrating what direction re-

search/tooling is heading.

2) Researchers interested in developing new col-

laborative modeling approaches and seeing a)

what areas are gaining the most traction, and b) if

the trends are appropriate or if new topics should

be addressed.

• Intervention: Techniques intended to facilitate

coordination, communication, and collaboration

(3C) among those employing software modeling

and MDE.

• Comparison: Not applicable. To identify con-

cepts and trends, we consider all techniques.

• Outcomes: Research and approaches that facili-

tate or support collaboration in MDE.

• Context: Published academic research.

This allowed us to form the research questions of

this paper, which are a reﬁnement of the more general

question we phrased earlier:

1. What are the emerging concepts in all collabora-

tive modeling research articles?

2. What are the emerging concepts in collaborative

modeling research articles published from 2012

until 2017, inclusively?

3. What are the (changing) trends implied by con-

trasting these two sets of concepts?

3.2 Search Strategy

Our search strategy included all major online libraries

relevant to this domain (Kitchenham and Charters,

2007). This included IEEExplore, ACM Digital Li-

brary, Google Scholar, and ScienceDirect. We ad-

ditionally consider all proceedings from the “Inter-

national Workshop on Model-Driven Engineering on

and for the Cloud” and the “International Workshop

on Collaborative Modelling in MDE”. Lastly, we

include/cross-check our list with all primary articles

identiﬁed in the survey performed by Micro Franzago

et al. (Franzago et al., 2017) to help ensure article cov-

erage.

Our search string includes any entries contain-

ing the term “collaborative modeling” in the meta-

data. The search string further included articles

with metadata containing the words “collaboration”

or “collaborative” with the union of any of “MDE”,

“Model driven engineering”, “software modeling”,

and “domain speciﬁc modeling language”. We in-

cluded both spellings of “modeling” and “modelling”

in all terms, as recommended in the SLR guide-

lines (Kitchenham and Charters, 2007) when dis-

cussing word variations. While the exact query we

Emerging Concepts and Trends in Collaborative Modeling: A Survey

241

used varied depending on which library we were con-

sidering, it appeared roughly in the form of the fol-

lowing exact query from our IEEE library search,

(“collaborative modelling” OR “collaborative mod-

eling”) OR ((“Document Title”:collaboration OR

“Document Title”:collaborative) AND (“Document

Title”:MDE OR “Document Title”:“Model driven

engineering” OR “Document Title”:“software mod-

eling” OR “Document Title”:“software modelling”

OR “Document Title”:“domain speciﬁc modeling

language” OR “Document Title”:“domain speciﬁc

modelling language” )).

3.3 Study Selection Criteria

Our study selection criteria applied to the search re-

sults ﬁrst ensures that each primary study describes

collaborative software modeling tooling and research.

Our initial criteria involves us including works from

all years. From that list, we identiﬁed those published

over the ﬁve year period of 2012 to 2017. We chose

not to include the current year, 2018, as 1) the year is

not complete, 2) it is possible that articles published

earlier this year may not have been disseminated or

included in libraries yet. Thus, in order to not make

the overly-bold claim of including 2018 and present

potentially skewed data, we do not include any arti-

cles from 2018. We excluded any non-English arti-

cles. We excluded any existing surveys, but consid-

ered their content when evaluating our search terms

and results.

3.4 Study Selection Procedure

The author of this paper performed the study extrac-

tion and collection using the search terms. Inclu-

sion/exclusion was performed through careful review

of the abstract and paper.

3.5 Study Quality Assessment

Procedure

Our inclusion criteria were purposely less strict than

the Franzago et al. survey, as we 1) did not require

our articles to include study evaluation and 2) in-

cluded Master’s and Ph.D. theses. We argue these

are relevant articles for our context and intention of

considering trends and directions. That is, evalua-

tion/validation and being thesis work should not be

a reason to exclude articles when considering emerg-

ing topics and trends. Our exclusion criteria was

any work not addressing methodologies and tech-

niques for collaborative software modeling as we de-

scribed in our PICOC intervention. Some exam-

ples of work we found using our query that we ex-

cluded includes primary studies on domain languages

for generating collaborative software; secondary stud-

ies, which we described earlier; research dealing with

hardware/circuity employing software instead of soft-

ware development, and others.

3.6 Data Extraction Strategy and

Synthesis

To discover emerging concepts in collaborative mod-

eling, we decided to analyze articles’ abstracts. We

decided on abstracts as they are intended to cover an

article’s important concepts and keywords, and, as ar-

gued by Paraschiv et al., form a good basis and data

for domain analysis (Paraschiv et al., 2015). We be-

gan by entering all the articles’ metadata into our bib-

liography manager and verifying the entries manually.

We then made a copy of that bibliography manager

list, and updated that copy to include entries from

2012 until 2017 only. For both lists, we extracted

out the abstracts into separate text ﬁles. We then

passed these text ﬁles to word cloud generator soft-

ware

, which generates both word clouds and associ-

ated word counts. Word clouds are not perfect, how-

ever, so we took some additional steps (Harris, 2011).

Our software allowed us to remove words from the

word cloud and counts. To help eliminate noise, we

ﬁltered out context-establishing words such as “col-

laborative”, “modeling”, “software”, “paper”, “engi-

neering”, and others. We additionally ﬁltered out

non-conceptual words such as “can”, “use”, “work”,

“present”, and others. We counted the amount of top

words in distinct abstracts to present as data. For re-

producibility, readers can ﬁnd the bibliography ﬁles

and abstracts, the list of included words and their

counts, and the removed words and their counts on

our associated web repository

. For the sake of

brevity and paper-length constraints, we list all our

primary studies on that web repository rather than in

this paper.

3.7 Results of Protocol

We breakdown how many papers we found in each

of the most prominent libraries in Table 1. Using our

protocol, we discovered 58 articles published in 2012-

2017, 45 articles published before 2012, and 103 ar-

ticles overall. All of the metadata for these articles

can be accessed on our associated web repository as

further reading.

https://www.wordclouds.com/

https://sc.lib.miamioh.edu/handle/2374.MIA/6283

MODELSWARD 2019 - 7th International Conference on Model-Driven Engineering and Software Development

242

Table 1: Primary Studies Found in Main Libraries.

Library # of Primary Studies

IEEE 137

ACM 157

Science Direct 45

Google Scholar 50

4 SUMMARY OF TREND DATA

We present the quantitative results of our ﬁndings here

and discuss them in the next section. While word

clouds may make it difﬁcult to search for words of

interest to readers, we present our complete lists as

further reading on our associated web page.

4.1 Articles from 2012 - 2017

We provide a word cloud for collaborative modeling

articles from 2012 until 2017 in Figure 1. Top topi-

cal words include “different”, “conﬂicts”, “control”,

“management”, “detection”, “requirements”, “qual-

ity”, “stakeholders”, “conﬂict”, “version”. We illus-

trate their prevalence in Table 2. The list is sorted

by the ﬁrst column, which indicates the total number

of word occurrences in all of the abstracts’ text. The

second column illustrates the amount of distinct ab-

stracts the word was found, and the third column indi-

cates the percentage, rounded to two decimal places,

of article abstracts the word appeared in out of the 58

articles.

4.2 All Articles

Readers can ﬁnd a word cloud representing all the

collaborative modeling articles until 2017 in Figure

2. Prevalent topical words found in the abstracts

in all articles include “UML”, “different”, “control”,

“management”, “version”, “integrated”, “merging”,

“CASE”, “architecture”. We list these words in Ta-

ble 3, sorted by the ﬁrst column indicating the to-

tal number of instances of the words in all abstracts.

The second column identiﬁes the number of distinct

articles with which the word appears, and the ﬁnal

column presents the percentage of all 103 article ab-

stracts that contain the word.

5 CHANGING TRENDS

To discuss changing trends, we begin by contrasting

the two word clouds and total counts. From a poten-

tial bias perspective, it is important to consider there

are 45 articles before 2012 and 58 articles after. Thus,

it is possible concepts in abstracts from 2012 until

2017 could be inﬂated. We could have chosen a dif-

ferent end year, however, we were interested in isolat-

ing recent, 5-year, research to provide a snapshot of

the most recent complete half decade. The abstracts’

sizes present a minor bias/threat to validity as the ab-

stracts we found ranged from 150-250 words, with

the majority of abstracts being in the 250 upper limit

range. We identify more threats to validity later in this

paper.

We illustrate some example comparisons in Fig-

ure 3 using a 100% stacked column chart. Speciﬁ-

cally, we showcase words that were prominent on one

list and not the other. We also include one word, “ver-

sion”, which was about an equal split, as a middle

baseline and interesting example. The darker shade

represents the percentage of instances of each respec-

tive word we found in article abstracts from articles

prior to the year 2012. The lighter shade represents

the percentage of instances of the respective word

found in article abstracts in articles from 2012-2017.

Therefore, columns that are darker represent con-

cepts that were more prevalent before 2012. Lighter

columns represent concepts that emerged and gained

prevalence in the years of 2012 to 2017.

5.1 Discussion

Unsurprisingly, we note that CASE model collabo-

ration research was conducted exclusively prior to

2012, which can be explained by the evolution of

CASE into object oriented approaches. We also note

that UML speciﬁc research was much more predom-

inate prior to 2012. Franzago et al. state that there

is “a prominence of UML-based approaches” (Fran-

zago et al., 2017), however, we take that one step fur-

ther and note that “prominence” is outdated as we dis-

covered the recent research focus is becoming more

language agnostic. They also note that there are rel-

atively few approaches that consider the interaction

among synchronous and asynchronous collaboration.

Our data further indicates this trend is likely to con-

tinue as the majority of both synchronous and asyn-

chronous research occurred prior to 2012. Speciﬁ-

cally, we found 20 instances of either topic prior to

2012 and only 5 instances of those topics in 2012-

2017 articles.

On the other end of the spectrum, we observe a

number of collaborative modeling research concepts

emerging in the last ﬁve complete years. While the

prevalence of “version” and “different” were roughly

the same before and after 2012, it seems the notions of

“conﬂicts” and “inconsistencies” have become more

Emerging Concepts and Trends in Collaborative Modeling: A Survey

243

Table 2: Top Words in Abstracts from Articles in 2012-2017.

Word # of Total Instances # of Articles Word is Found % of 58 Articles

Different 23 19 33%

Conﬂicts 22 11 19%

Control 21 11 19%

Management 14 11 19%

Detection 13 6 10%

Requirements 12 8 14%

Quality 12 9 16%

Stakeholders 11 10 17%

Conﬂict 11 6 10%

Version 10 9 16%

Figure 1: Emerging Concepts in Abstracts from Articles Published 2012 to 2017.

MODELSWARD 2019 - 7th International Conference on Model-Driven Engineering and Software Development

244

Figure 2: Emerging Concepts in Abstracts from All Articles.

Table 3: Top Words in Abstracts from All Articles.

Word # of Total Instances # of Articles Word is Found % of 103 Articles

UML 42 21 20%

Different 39 30 29%

Control 38 22 21%

Management 35 21 20%

Conﬂicts 34 17 17%

Version 23 16 16%

Integrated 17 12 12%

Merging 17 10 10%

Case 17 6 6%

Architecture 16 11 11%

Emerging Concepts and Trends in Collaborative Modeling: A Survey

245

Figure 3: A 100% Stacked Column Chart of Abstract Word Occurrence Before 2012 and After.

of a focus in recent years. This can potentially be

explained by the need for conﬂict resolution arising

from the advent of the more advanced systems being

built by organizations and the increasingly intricate

and necessary collaborations experienced by model-

ers. It also coincides with the increase in model com-

parison research (Stephan and Cordy, 2012; Stephan

and Cordy, 2013), which involves detecting differ-

ences and conﬂicts among a set of models. Our

trend chart also demonstrates an increase in the need

for assessing and establishing “quality” in collabo-

rative modeling. This corresponds with the increase

in research and industrial desire for quality evalua-

tion in software modeling in general (Mohagheghi

and Aagedal, 2007; Giraldo et al., 2016). Security,

an aspect of quality, demonstrates the same trend,

as “access” and “control” concepts are more preva-

lent in the post 2011 research. This is not surpris-

ing given the high security focus that many organiza-

tions are employing due to very public software se-

curity breaches in the media (Sen, 2018). Another

trending concept, not shown in our chart, is “telecom-

munications”. All research referencing telecommu-

nications was found to be published exclusively in

2012 or later. An interesting observation is the shift

of focus from “developers” to “stakeholders”. While

stakeholders is an all-encompassing term that po-

tentially includes developers, this may also reﬂect

the stakeholder/customer driven Agile software de-

velopment methodology growth (Abrahamsson et al.,

2002) where methods and processes are focused on

all stakeholders, rather than just the developers.

5.1.1 Limitations and Threats to Validity

One aspect we did not consider necessary was to em-

ploy a taxonomy to help characterize/categorize and

index the results. While Franzago’s study included

one (Franzago et al., 2017), it made more sense for

their purposes of a mapping study intended to iden-

tify “white spots” to warrant further research. That

is, our study was more reﬂective, and theirs was more

prescriptive. Secondly, we limited our paper to word

clouds based on abstracts, which limited the paper

to a very abstract level. That was our intention, but

that does carry with it some potential threats to our

trend data. While we contend it is useful for those at

the managerial level and practitioners, it could be less

helpful for the latter as our data is abstract. It may be

a technically implemented survey, but we conducted

it in an original, and temporally-focused, manner and

believe it has value for the reader and collaborative

modeling community. At the very least, readers can 1)

access our primary study metadata on our repository

as a good source of information, 2) beneﬁt from an

abstract view of trends and contrast of different time

periods. Lastly, we acknowledged that we purposely

left out 2018 data for the reasons we discussed in our

study selection section.

MODELSWARD 2019 - 7th International Conference on Model-Driven Engineering and Software Development

246

6 CONCLUSION

Based on the sheer number of articles we found in

our systematic review from 2012 to 2017 compared to

all articles, it is clear collaborative modeling research

is very topical and on the rise. This is additionally

supported by the emergence of dedicated workshops

such as the “International Workshop on Collabora-

tive Modelling in MDE” and explicit sessions dedi-

cated to modeling at conferences. We learned that,

while the importance of model version control has re-

mained consistent, there has been a relatively recent

inﬂux and focus on detecting and dealing with con-

ﬂicts and inconsistencies. Additionally, the quality of

models created through collaborative modeling, and

security access in the collaboration process are be-

coming more paramount. It is likely that the research

results from this relatively recent focus will permeate

into collaborative tooling, more so than it already has.

While we were concerned with what research was

being completed and published, a future goal of both

practitioners and researchers must be to better com-

municate and work together more so that the focus

and direction of research is coming from practition-

ers. In the majority of the research we encountered,

it appears that researchers were tackling important

problems, but there was no evidence nor support that

these were the areas most interesting and desirably

to practitioners. The general interests of practition-

ers and researchers align in theory, and thus should be

more connected. It is our hope that this article’s at-a-

glance overview of emerging trends in collaborative

modeling research may inspire practitioners to reach

out to researchers to let them know if these trends cor-

respond to their interests or if there is something else

researchers should be focusing on to better help serve

the software modeling community.

REFERENCES

Abrahamsson, P., Salo, O., Ronkainen, J., and Warsta, J.

(2002). Agile software development methods: Review

and analysis. Technical Report VTT Publications 478,

VTT Technical Research Centre of Finland.

Cui, W., Wu, Y., Liu, S., Wei, F., Zhou, M. X., and Qu, H.

(2010). Context preserving dynamic word cloud visu-

alization. In 2010 IEEE Paciﬁc Visualization Sympo-

sium (PaciﬁcVis), pages 121–128.

Franzago, M., Di Ruscio, D., Malavolta, I., and Muccini,

H. (2017). Collaborative model-driven software en-

gineering: a classiﬁcation framework and a research

map. IEEE Transactions on Software Engineering.

Giraldo, F. D., Espana, S., Pastor, O., and Giraldo, W. J.

(2016). Considerations about quality in model-driven

engineering. Software Quality Journal, pages 1–66.

Harris, J. (2011). Word clouds considered harmful. Nieman

Journalism Lab.

Hutchinson, J., Rounceﬁeld, M., and Whittle, J. (2011).

Model-driven engineering practices in industry. In

Proceedings of the 33rd International Conference on

Software Engineering, pages 633–642. ACM.

Kitchenham, B. and Charters, S. (2007). Guidelines for per-

forming systematic literature reviews in software en-

gineering. Technical Report Technical Report EBSE-

2007-01, School of Computer Science and Mathemat-

ics, Keele University.

McIlraith, S. A., Son, T. C., and Zeng, H. (2001). Semantic

web services. IEEE intelligent systems, 16(2):46–53.

Mellor, S. J., Clark, T., and Futagami, T. (2003). Model-

driven development: guest editors’ introduction. IEEE

software, 20(5):14–18.

Mohagheghi, P. and Aagedal, J. (2007). Evaluating quality

in model-driven engineering. In International Work-

shop on Modeling in Software Engineering (MISE’07:

ICSE Workshop 2007), pages 1–6.

Ober, I., Gokhale, A., Georg, G., Felderer, M., Bruel, J.-M.,

and Lugato, D. (2012). Mdhpcl 2012 workshop sum-

mary. In Proceedings of the 1st International Work-

shop on Model-Driven Engineering for High Perfor-

mance and CLoud Computing, MDHPCL ’12, pages

1:1–1:2, New York, NY, USA. ACM.

Paraschiv, I. C., Dascalu, M., Trausan-Matu, S., and Dessus,

P. (2015). A semantic approach to analyze scientiﬁc

paper abstracts. In The International Scientiﬁc Con-

ference eLearning and Software for Education, vol-

ume 1, pages 3–9.

Petticrew, M. and Roberts, H. (2008). Systematic reviews in

the social sciences: A practical guide. John Wiley &

Sons.

Portillo-Rodr

ıguez, J., Vizca

ıno, A., Piattini, M., and

Beecham, S. (2012). Tools used in global software

engineering: A systematic mapping review. Informa-

tion and Software Technology, 54(7):663–685.

Renger, M., Kolfschoten, G. L., and De Vreede, G.-J.

(2008). Challenges in collaborative modelling: a liter-

ature review and research agenda. International Jour-

nal of Simulation and Process Modelling, 4(3-4):248–

263.

Rocha, R. G., Costa, C., Rodrigues, C., Ribeiro de Azevedo,

R., Junior, I. H., Meira, S., and Prikladnicki, R.

(2011). Collaboration models in distributed software

development: A systematic review. CLEI Electronic

Journal, 14(2):1–2.

Sen, R. (2018). Challenges to cybersecurity: Current state

of affairs. Communications of the Association for In-

formation Systems, 43(1):2.

Stephan, M. and Cordy, J. R. (2012). A Survey of Meth-

ods and Applications of Model Comparison. Techni-

cal Report 2011-582, Queen’s University. Revision 3.

Stephan, M. and Cordy, J. R. (2013). A Survey of Model

Comparison Approaches and Applications. In In-

ternational Conference on Model-Driven Engineering

and Software Development (Modelsward), pages 265–

277. SCITEPRESS.

Emerging Concepts and Trends in Collaborative Modeling: A Survey

247