A Systematic Mapping Study on Domain-Specific Languages

Abstract

Domain-Specific Languages (DSLs) offer substantial gains in expressiveness and ease of use compared with general purpose languages. This way, DSLs have gained significant attention in industry and academy, as can be seen by the increased number of related publications in key conferences and journals. This paper aims to provide a broad view of the DSL research field by performing a Systematic Mapping Study. Adopting a detailed search strategy, 4450 studies were initially identified, and, after filtering, 1440 primary studies were selected and categorized using a particular classification scheme. So, this work presents the most popular application domains where DSLs have been applied, identifies different tools for handling DSLs, including language workbenches, and enumerates several techniques, methods and/or processes for dealing with DSLs.

Full text

A Systematic Mapping Study on Domain-Specific Languages
Leandro Marques do Nascimento1,2, Daniel Leite Viana1, Paulo A. M. Silveira Neto1,3,
Dhiego A. O. Martins1, Vinicius Cardoso Garcia1, Silvio R. L. Meira1
1 Informatics Center, Federal University of Pernambuco (UFPE)
Recife, Brazil
2 Department of Informatics, Federal Rural University of Pernambuco (UFRPE)
Recife, Brazil
3 Department of Informatics, Federal Rural University of Pernambuco (UFRPE)
Serra Talhada, Brazil
{lmn2, dlv2, pamsn, daom, vcg, srlm}@cin.ufpe.br
Abstract—Domain-Specific Languages (DSLs) offer substantial
gains in expressiveness and ease of use compared with general
purpose languages. This way, DSLs have gained significant
attention in industry and academy, as can be seen by the
increased number of related publications in key conferences
and journals. This paper aims to provide a broad view of the
DSL research field by performing a Systematic Mapping
Study. Adopting a detailed search strategy, 4450 studies were
initially identified, and, after filtering, 1440 primary studies
were selected and categorized using a particular classification
scheme. So, this work presents the most popular application
domains where DSLs have been applied, identifies different
tools for handling DSLs, including language workbenches, and
enumerates several techniques, methods and/or processes for
dealing with DSLs.
Keywords: Domain-specific languages; systematic mapping
study; programming languages; mini languages; little languages.
I. INTRODUCTION
Software systems are built upon computer languages or
better called programming languages. A programming
language is a notation for expressing computations
(algorithms) in both machine and human readable form.
Appropriate programming languages and tools may
drastically reduce the cost of building new applications as
well as maintaining existing ones [1]. For humans, it would
be easier to write computer programs if a natural language
could be used, such as English or Portuguese, for instance.
However, computer languages must follow a rigid
predefined structure, with a specific grammar and syntax,
and to learn this structure is not so easy for many people,
taking significant ti      
kind of language.
In the context of programming languages, a Domain-
Specific Language (DSL) is a language that provides
constructs and notations tailored toward a particular
application domain [2]. Usually, DSLs are small, more
declarative than imperative, and more attractive than
General-Purpose Languages (GPL) for their particular
application domain due to easier program understanding,
reduced semantic distance between the problem and the
program, and enhanced productivity. Some well-known
examples of DSLs are BNF (syntax definition), HTML
(hypertext markup), SQL (database queries), and VHDL
(hardware design).
DSLs trade generality for expressiveness in a limited
domain, and this can bring several benefits to software
engineering. However, these benefits do not come for free.
The cost of DSL design, development and maintenance has
to be taken into account. Without appropriate methodologies
and/or tools these costs can be higher than savings. Although
DSLs have been developed from the beginning of computer
science (an early example is APT, a DSL for numerical
control of machine tools developed back in the 1950s at MIT
[3]), many unanswered questions remain regarding when and
how to develop a DSL.
Therefore, this paper presents a systematic mapping
study in order to better understand the DSL research field,
through synthesizing evidence to suggest important
implications for practice, as well as identifying research
trends, open issues, and areas for improvement. A Mapping
    [4] is an evidence-
based approach, applied in order to provide an overview of a
research area, and to identify the quantity and type of
research and results available within it. Hence, the goal of
this investigation is to identify, evaluate, and synthesize
state-of-the-art domain-specific programming practices in
gathering evidence of what has been achieved so far in this
discipline. We are also interested in cataloging which are the
domains that have taken advantage of using DSLs. This way,
researchers and/or practitioners may know which DSLs have
been applied to a particular domain and then reuse or adapt it
for any other specific needs. This systematic mapping
process was conducted from November, 2011 to April, 2012.
The remainder of this paper is organized as follows:
Section 2 presents the related work. In Section 3, the
research methodology used in this paper is described
including the research questions, the search strategy and the
classification scheme. Section 4 reports the main findings. In
Section 5, the threats to validity are shown, and at last,
Section 6 draws some conclusions and provides
recommendations for further research on this topic.
II. RELATED WORK
The literature on DSLs provides a large number of
studies, regarding both general and specific issues, as will be
179Copyright (c) IARIA, 2012. ISBN: 978-1-61208-230-1
ICSEA 2012 : The Seventh International Conference on Software Engineering Advances

discussed later in this paper. However, a general search for
(“mapping study” OR “systematic literature review”) AND
“domain-specific languages” in well-known search engines
have shown that no publication have tried to address the
issues of this research field using specifically the MS
approach. Actually, many papers presented the state-of-the-
art in this field using other approaches than a MS and they
are next described as related work.
One of the first published papers to coin the concept of a
DSL is from 1965 [5]. It presents a family of unimplemented
computing languages that is intended to span differences of a
given application area by a unified framework.
In the 1980s, Bentley [6] tried to summarize the concept
of the so called Little Languages. The paper describes
examples of small languages that could be developed with
the technology available back there, e.g. COBOL and
FORTRAN.
In a paper from 2000, Deursen et al. [7] list a selection of
75 key publications in the area. It discusses terminology,
risks and benefits, examples of domain-specific languages,
design methodologies, and implementation techniques.
In a more recent work from 2005, Mernik et al. [2] try to
   When and How to Develop Domain-
Specific Languages?       
developed until then for different domains. The work
identified five DSL development phases: decision, analysis,
design, implementation, and deployment, and then relates the
listed DSLs with their development phases. At last, the work
enumerates domain analysis tools and language development
systems, giving a full view of the open issues in the area.
One of the most recent related work that could be
identified is [8], from 2011. It compares four different
approaches for DSL implementation: ANTLR, Ruby,
Stratego and Converge. From their comparative study, it was
observed that each approach has its merits and demerits and
there is no single approach that would apply to all scenarios.
The work does not mention directly the use of language
workbenches.
Indeed, we believe our study states current and relevant
information on research topics that can complement others
previously published. By current, we mean that, as the
number of studies published has increased rapidly, as shown
in Figure 2, it justifies the need of more up to date empirical
research in this area to contribute to the community
investigations. Moreover, applying a MS approach to map
out the research area of DSL gives us a full overview of what
is being done and what is lacking attention from academia/
industry, as well as allow future extensions and replications.
III. RESEARCH METHODOLOGY
The experimental software engineering community is
working towards the definition of a standard processes for
conducting literature reviews. There are mainly two different
approaches to be cited: Systematic Literature Reviews (SR)
and Systematic Mapping Studies (MS) [9]. While a SR is a
mean of identifying, evaluating, interpreting and comparing
all available research relevant to a particular question [9], a

to answer detailed research questions [4]. A MS comprises
the analysis of primary studies that investigate aspects
related to predefined research questions, aiming at
integrating and synthesizing evidence to support or refute
particular research hypotheses.
In this study, we merged ideas from Petersen et al. [4]
with some good practices defined in the guidelines proposed
by Kitchenham and Charters [9], such as the protocol
definition. Therefore, we could apply a process for a
mapping study, including best practices for conducting
systematic reviews, making the best use of both techniques.
A MS is basically performed in three phases. All phases
are detailed in following sections: 1) Definition of the
protocol, which comprises the research questions and the
search strategy. This phase is commonly used in systematic
reviews. 2) Conducting the study with screening of relevant
papers. During this phase, a classification scheme is used. 3)
Keywording relevant topics, data extraction and systematic
mapping.
A. Research Questions
This mapping study intends to identify relevant
publications about Domain-Specific Languages,
understanding how they can be created and which ones have
been created so far. In addition, this study tries to enumerate
the domains in which DSLs have been applied, which
knowledge is necessary from the domain experts to start
using the language, and so forth which are the open issues of
the whole research field.
In summary the main research question of this study is:
In which manner are Domain Specific Languages (DSLs)
being created, used and maintained?
B. Research Sub-questions
Moreover, in order to make the mapping study main
objective more clear and repeatable, some research sub-
questions are defined, as following:
Q1. Which techniques, methods and/or processes are
used while working with DSLs, i.e. creation,
application, evolution and extension of DSLs?
Q2. Which DSLs have been created and are available for
use or are described in some type of publication?
Q3. In which domains are these DSLs being used?
Q4. Which tools are used for the development and usage
of DSLs and how such tools support those activities?
C. Search Strategy, Data Sources and Studies Selection
According to our research questions and in order to
increase the coverage of our search, we decided to use the
following search string, which brings only general terms
grouped by an OR clause:
"domain-specific language" OR "domain-specific
modeling language" OR "generative programming"
Therefore, instead of restricting our search items with
other keywords, we understand that any work that mentions
one of the three items listed is going to be returned by the
search engine anyway. Although the number of manuscripts
returned could increase considerably, few or even no
relevant studies would be left over. Indeed, experts in the
180Copyright (c) IARIA, 2012. ISBN: 978-1-61208-230-1
ICSEA 2012 : The Seventh International Conference on Software Engineering Advances

DSL research field may say there are other related terms,
little/small language Architecture Description
Language ng those terms in our
automatic search, we decided to look for those terms in the
manual search and snow-balling process (which follows up
the reference list of each selected manuscript), since papers
          domain-
specific language        
during a fine-grained and non-automatic search process.
    domain-specific modeling
language  generative programming  
noticed that these terms are extremely related to the research
field just by checking at the most relevant papers according
to the search engines relevance ordering.
The study was conducted using automatic and manual
search. We did not establish any inferior year-limit. For
automatic search, six search engines and digital databases of
scientific sources were used: ACM Digital Library,
IEEEXplore, SpringerLink, Science Direct, Scopus and
Engineering Village (also known as El Compendex).
Besides, the manual search includes the most important
international, peer-reviewed journals published by Elsevier,
IEEE, ACM and Springer, and 26 different conferences.
After performing the automatic and manual search, a
total of 4450 papers were identified, 93 of them from manual
search. During manual search, a snow-balling process was
done. Next, the studies were submitted to the inclusion and
exclusion criteria, as we detail in the following section.
The studies selection involved a screening process
composed of three filters, in order to select the most suitable
results, since the likelihood of retrieving not adequate studies
might be high. Figure 1 details each filter.
Regarding the inclusion criteria, the studies were
submitted to the following conditions:
 
regarding Domain Specific Languages, Domain
Specific Modeling Languages and/or Generative
Programming. No date filtering was applied.
 While verifying if a given article may be included in
our study, we can check if it is possible to answer

o Is it a DSL or DSML?
o Is it a technique, method or process for
handling DSLs/DSMLs?
o Is it a tool (language workbench) for handling
DSLs/DSMLs?
o Is it any type of philosophical paper that
discusses concepts of DSLs, DSMLs and/or
any related generative programming
technique?
Considering the exclusion criteria, the studies were
submitted to the following conditions:
 Articles not written in English.
 Literature that was only available in the form of
abstracts or Powerpoint presentations. Posters, short
papers (less than 2 pages) and invited conference
talks with no relevant results can be excluded.
 Articles in press, journals and conferences
editorials/reviews can also be excluded.
 Duplicated and/or incomplete studies.
Figure 1. Stages of the selection process and the corresponding number of
papers.
After performing the selection process, some results can
be seen in Figure 2 which shows the distribution of the
primary studies, considering the publication year. The Figure
2 clearly gives us the impression that many correlated areas
in software engineering and computer science in general are
taking more and more advantage of DSLs in practice, as we
can check by looking at the growth curve.
Figure 2. Distribution of studies by their publication years after 3rd filter.
We were able to identify the most common locals of
publication. Conferences such ICSE (International
Conference on Software Engineering), OOPSLA (Object-
Oriented Programming, Systems, Languages & Applications
Conference) and GPCE (Generative Programming and
Component Engineering Conference) had the highest
number of studies published. Similarly, the most popular
journals were ACM SIGPLAN Notices, IEEE Software and
ENTCS (Electronic Notes in Theoretical Computer). We
catalogued manuscripts from 548 different sources (418
conferences and 130 journals).
D. Classifying Selected Studies
Our classification scheme assembled three facets. Facet
one lists the classes of research based on [4]: Validation
Research, Evaluation Research, Solution Proposal,
Philosophical Papers, Opinion Papers, Experience Papers.
Details of each class of research can be found on [4]. The
two others are directly related to our research questions.
Filters Activities N. of papers
1
2
3
Identify relevant studies using
defined search terms through
manual and automatic search
Exclude studies based on
exclusion criteria for
removing duplicated studies
Exclude studies by applying
exclusion criteria on title,
abstract and keywords.
N = 4450
N = 2688
N = 1440
12112 2 321452427274
19 31 25 32 35 45
59 70
100
138
172
221 225
216
0
50
100
150
200
250
1966
1967
1971
1973
1975
1976
1977
1986
1988
1989
1990
1991
1992
1993
1994
1995
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
Distribution of primary studies
181Copyright (c) IARIA, 2012. ISBN: 978-1-61208-230-1
ICSEA 2012 : The Seventh International Conference on Software Engineering Advances

Facet 2  DSL Research Type  considered in our study
is directly related to the research sub-questions Q1, Q2 and
Q4. We tried to identify studies that report specifically the
usage of a given DSL to solve a problem and also studies
that report any kind of technique, method or process to
handle DSLs, i.e., create, evolve, integrate, debug. What is
more, we tried to enumerate what tools have been used to
apply those techniques, methods and/or processes. TABLE I
presents the details of Facet 2. Some concepts of this facet
are based on [10].
TABLE I. FACET 2  DSL RESEARCH TYPE.
1. ADL
Architecture Description Languages (ADLs) are
aimed at the specification of high level system
architectures, described in terms of components
and connectors.
2. DSAL
A Domain-Specific Aspect Language combines
benefits from DSLs and Aspect-Oriented
Programming (AOP). It is a aspect language
tailored to a specific domain.
3. DSML
A domain-specific modeling language is a
special type of DSL that can be used for
modeling domain-specific systems. The concept
of a DSML comes originally from the adaptation
of UML to specific domains.
4. External DSL
A completely separate language, for which you
write a full parser, usually using a parser
generator.
5. Internal DSL
An internal (or embedded) DSL is an idiomatic
way of using a general-purpose language.
6. Method or Process
Any type of generic solution for a class of
problems which usually involves technical and
non-technical aspects. A method/process
involves a set of steps to be performed in order
to make it repeatable for anyone to try using it.
A method/process may use a group of
techniques which combined represent a generic
solution for a class of problems.
7. Technique
Any type of solution for a specific problem. For
example: a technique to generate Java code
based on C# input; a technique for teaching how
to create parsers, a technique to analyze model
coupling.
8. Tools
Any type of software engineering tool used for
handling DSLs.
Facet 3 addresses the domains in which DSL techniques
are somehow applied and is directly related to Q3. Inspired
by previous publications that tried to do the same [2], [7], we
identified many different domains ranging from
bioinformatics to robotics and control systems, for example.
We were able to enumerate 30 different domains. Among
them, we selected the top 15 most referenced domains to be
used as facet in this study. Since the final number of papers
included in our study was quite large (1440), some domains
were mentioned few times (1 or 2), then those ones are not
considered to our classification. TABLE II displays Facet 3.
It is important to notice that none of the three facets are
exclusive, it means, a paper may be classified in two or more
categories of any of the three facets. For example, a paper
may be categorized as a Solution Proposal and Validation
Research, as a DSML and a Tool and also with the domains
of Web and Control Systems.
TABLE II. FACET 3  DOMAINS
1. Web
Every study that uses any type of web
technology
2. Embedded Systems
Hardware and software co-design
3. Low-level Software
Low-level programming, for instance,
operating systems, device drivers, etc.
4. Control Systems
Any type of control systems, for example:
flight control, automation systems, etc.
5. Parallel Computing
High-performance computing, multithreaded
programming
6. Simulation
Any type of simulation software
7. Data Intensive Apps
Studies that present ways of handling
databases using DSL techniques
8. Real-time Systems
Systems where the time is a crucial variable
9. Security
Studies that handle security issues such as
intrusion detection, access control, etc.
10. Dynamic Systems
A type of software system that can adapt to
the context it is immersed
11. Visual Language
Apart from textual languages, this type of
study describes a DSL with visual appealing
12. Testing
DSLs applied to the software engineering
discipline of testing
13. Education
Any type of publication that mentions
education as the primary goal, e.g. as in [11]
14. Network
DSLs for manipulating computer networks
and/or distributed systems issues
15. Others
In this category, we gathered domains with at
most 5 publications, covering several
divergent topics, such as Chemistry,
Geometry and Engineering, among others
In addition, it is important to highlight that TABLE II is
missing some important domains due the total amount of
manuscripts included in this study. Hereby, we cite one
sample publication of these domains that were left over:
healthcare [12], pervasive computing [13], graphics [14],
cloud/grid computing [15], robotics [16], ontology [17],
games [18], multi-agent systems [19], requirements
engineering [20], bioinformatics [21], mobile apps [22],
multimedia [23], user interface [24], hardware description
[25], automation [26].
IV. MAIN FINDINGS
In this section, each topic presents the findings of a
research sub-question, highlighting evidences gathered from
the data extraction process. These results populate the
classification scheme, which evolves while doing the data
extraction. It is important to mention that this study is not
going to enumerate all the references we found, as it makes
no sense at all to list 1440 references. Instead, we are going
to choose sample references to demonstrate our results.
Our first results are shown in Figure 3, which presents
the distribution of papers according to Facet 1  Classes of
research. As can be seen, there is a majority number of
Solution Proposals, which indicates that there are many
proposals yet to be validated. The number of Validation and
Evaluation Research together represents about one third of
those proposals, which means that a representative number
of proposals are somehow tested in industry and/or academy.
182Copyright (c) IARIA, 2012. ISBN: 978-1-61208-230-1
ICSEA 2012 : The Seventh International Conference on Software Engineering Advances

Figure 3. Distribution of papers by classes of research.
A. Techniques, Methods and/or Processes for Handling
DSLs
Several techniques, methods and/or processes could be
found during the execution of this mapping study. Methods
for software construction using generative techniques are not
new as we can see in [27], although they did not directly
mention the construction of DSLs.
At an abstract level, a language is a means of
communication; in the case of computing that
communication is generally between a human and a
machine. In order to be usable, a language needs to have a
way that participants can share communications (syntax) and
an agreed shared meaning (semantics). Languages may form
parts of larger languages (e.g. the sub-part of English used
only in computing could be detached and reattached to the
main language); they may be parameterisable (e.g. American
and British English can be seen as variations on the single,
abstract, language English); they may have variable syntaxes
(e.g. Serbian is written in both the Cyrillic and Latin
alphabets); and so on [28].
One of the processes for handling DSL catalogued by this
mapping study is called Language Factories [28]. Language
Factories break languages down into components, including
the following parts:
 Abstract syntax: The single definition of its
Abstract Syntax Tree (AST).
 Concrete syntax(es) and syntactic mapping: A
definition of its concrete syntax(es) specified as e.g.
a context free grammar, and a mapping from that
concrete syntax to the abstract syntax.
 Semantic aspect(s): Each semantic aspect defines (a
possibly incomplete part of the) semantics. Semantic
aspects may overlap with each other (e.g. an
operational and denotational semantics) or describe
completely different elements of the semantics (e.g.
semantics of language types and semantics for text
editors supporting tool-tips).
 Constraints: Describes constraints on how the
language can be composed with others (both in
terms of what the component provides, and what it
requires of other components).
These parts of language development could help us in
citing the findings of this study. The development of formal
DSLs contains concepts of metamodels or grammars
(syntax) [29], [30], context conditions (static analysis and
quality assurance) as well as possibilities to define the
semantics of a language [31]. Many references highlight
techniques directly related to compiler construction [11],
[32], [33]. Along with the concept of DSL, we catalogued
some publications describing DSMLs and its peculiarities
[34]. Over the last few decades, DSLs have proven efficient
for mastering the complexities of software development
projects. The natural adaptation of DSLs to the model-driven
technologies has in turn established domain-
modeling languages (DSMLs) as vital tools for enhancing
design productivity.
A widespread approach to the design of a DSML is to
make use of the so-called profile mechanisms and to reuse
the UML metamodel as the base language. By extending
UML elements with stereotypes and their attributes, it is
possible to define new concepts to better represent elements
of a domain. Despite the ever increasing number of profiles
defined and successfully applied in many applications.
The technique of UML profile is mentioned in 21
publications of our catalogue, as for example, [3436]. We
noticed that many of those techniques are well supported by
tools, as we exemplify in the corresponding section.
We found quite a large number of techniques, methods
and/or processes as can be seen in Figure 4. These are some
examples of techniques for creating new DSLs: [3739]. A
total number of 160 publications mention some topic related
to DSL creation, 69 other publications mention DSML
creation and 53 mention embedded DSL creation.
Among different methods/processes for creating [40],
implementing [41] and evolving [42], [43] a DSL, one of the
methods that caught attention was the one that mentions
directly the concept of Language-Oriented Programming
[44] or even DSL oriented software engineering. The
        
right domain specific tool for each problem, instead of some
universal tool coupled with a way of working that tries to
wrap it so that it becomes usable in various contexts. The
primary meta-tool promoted in [44] is usage of high level,
strictly domain specific languages, based on formal concepts
used and widely understood by domain experts who may
have limited or no software engineering knowledge. This
concept of language-oriented programming is fully aligned
with other similar concept called Language Factories [28],
already mentioned.
Figure 4. Distribution of papers by DSL research type.
201
215
1142
29
28
105
0 200 400 600 800 1000 1200
Validation Research
Evaluation Research
Solution Proposal
Philosophical Paper
Opinion Paper
Experience Paper
Distribution of papers by classes of research
151
337
111
125
549
170
7
29
0 100 200 300 400 500 600
Tools
Technique
Method or Process
Internal DSL
External DSL
DSML
DSAL
ADL
Distribution of papers by DSL Research Type
183Copyright (c) IARIA, 2012. ISBN: 978-1-61208-230-1
ICSEA 2012 : The Seventh International Conference on Software Engineering Advances

Moreover, other relevant aspect identified in this study
involves DSL integration/composition, as shown in [4547].
Development of and tooling for a single DSL is well-studied,
but surprisingly little is known about the interplay of
different DSLs in a single system. Multiple DSLs are
required when moving from toy examples to real enterprise
applications. Methods and tool support are needed if multiple
DSL development is to succeed. One of these methods is
described in [48]. The method specifically tackles the
problem of overlapping concerns between different DSLs. It
has three steps: 1) Identification, 2) Specification, and 3)
Application. The purpose of the Identification step is to
uncover the overlaps between different languages and
identify connections among them. The Specification step
encodes these connections in a way that will make them
amenable to various analyses. The last step of the method is
Application where the encoded connections from the
previous two steps are used. The authors also provide tools
and case studies for using their method.
B. Domain-Specific Languages and their Respective
Domains
As can be seen in Figure 5, several DSLs were
catalogued according to their domain. We separated the
studies that simply report the usage of a DSL in two
categories: external DSL and internal (embedded DSL). For
each embedded DSL, we also identified in which technology
it was implemented. The most common technology in which
DSLs are embedded is Haskell with 46 concurrencies, as for
example in [49]. However many other host languages are
used, such as Java, C/C++, Ruby, Scala, SmallTalk, Python,
Prolog, XML and even some unpopular languages like
Clean, Galois, Dylan and Curry.
Figure 5. Distribution of papers highlighting top 15 domains.
Different types of DSLs have been identified other than
the ones we previously knew. We identified FSML, ADL
and DSAL.
A Framework-Specific Modeling Language (FSML) [50]
is a kind of Domain-Specific Modeling Language that is
used for modeling framework-based software. FSMLs enable
automated round-trip engineering over non-trivial model-to-
code mappings and thereby simplify the task of creating and
evolving framework-based applications.
An Architecture Description Language (ADL or ADSL)
[51]        
    directly   
       
ingredients of the conceptual architecture. Developers can
thus use the language to describe a system on the
architectural level.
A Domain-Specific Aspect Language (DSAL) [52] is a
custom language that allows special forms of crosscutting
concerns to be decomposed into modularized constructs.
Examples of domain-specific aspect languages include
languages for dealing with coordination concerns, object
marshaling concerns, and class graph traversal concerns.
Many different domains that make use of DSL could be
identified in our study. The most popular domain was the
horizontal domain of web applications, in which several
publications states the use of web services, and terms like
services composition, services orchestration and services
mash up are common. Figure 6 shows a full cross reference
view of the DSL research type and their respective domains.
In this context, web services composition refers to the
creation of new (web) services by combining functionalities
provided by existing ones. A number of domain-specific
languages for service composition have been proposed, with
consensus being formed around a process-oriented language
known as WS-BPEL (or BPEL). The kernel of BPEL
consists of simple communication primitives that may be
combined using control-flow constructs expressing
sequence, branching, parallelism, synchronization, etc.
Some examples of BPEL identified in this study: [5355].
C. Tools
Tools play an essential role in software engineering and it
is not different when we are talking doing language
engineering. Our study identified 151 manuscripts that are
related to DSL tools.
Some studies do not actually describe a new tool, but
discuss about other tools as in [56] or just make use of a set
of tools and report the experience as in [57]. Although, there
are few studies comparing DSL tools, we were able to
identify two of them as can be seen in [8], [58].
Observing the available publications, we could identify 3
subcategories of tools:
 Tools for using DSLs: this type of tool is actually
the more comfortable for the user once he/she is
supposed to be familiarized with the domain being
manipulated. No knowledge about language
engineering or domain engineering is necessary for
using this type of tool, as well as it is projected be
used by domain experts. A good example listed in
our study is the tool Scratch [59], appropriated for
introductory programming courses.
 Tools for DSL creation (specification): these are a
more intuitive way of creating compilers. At this
level, the tool is nothing more than a compiler of
compilers and, in the end of the process of DSL
creation, there will be no integration with other
software engineering tools (IDEs), pretty printer,
141
54
38
27
36
28
59
54
91
67
51
26
31
81
75
020 40 60 80 100 120 140 160
Web
Visual Language
Testing
Simulation
Security
Real-time Systems
Parallel Computing
Others
Network
Low-level Software
Embedded Systems
Education
Dynamic Systems
Data Intensive Apps
Control Systems
Distribution of papers highlighting top 15 domains
184Copyright (c) IARIA, 2012. ISBN: 978-1-61208-230-1
ICSEA 2012 : The Seventh International Conference on Software Engineering Advances

Figure 6. DSL Research Type VS Top 15 Domains.
code assistant and so on. An example of this type of
tool in our study is JTS (Jakarta Tool Suite) [60].
 Language workbench: these tools support DSL
creation not just in terms of parsing and code
generation but also in providing a better editing
experience for DSL users. In particular, language
workbenches let a DSL author create custom DSL
editors of similar power to modern IDEs. Language
workbenches are still in their early days, but if their
potential is realized, they could change the face of
programming [10]. Our study identified some
examples of language workbenches: XText [61],
MetaEdit+ [62], Spoofax [63], and MPS JetBrains
[64].
Another way of classifying tools is considering textual
and visual languages as described in [65]. Instead, we
decided to try other classification to highlight the real power
of language workbenches. Unfortunately, the number of
publications regarding language workbenches is still low.
However, some relevant studies have been published such as
[6466].
V. THREATS TO VALIDITY
There are some threats to the validity of our study. First
one is regarding our set of research questions. The set we
defined might not have covered the whole DSL research
field, mainly because language implementation in general
overlaps other several research fields, for example, model-
driven approaches. As we considered this as a feasible threat,
we had several discussion meetings and decided to use
questions as broad as possible. This way, we knew that the
number of primary studies would be bigger but there would
be a smaller chance of leaving any important study out of
this MS.
In addition, it is possible that we have not chosen the
most appropriate keywords. In general, several research
fields that use computer science as a mean to solve problems
also use DSLs to provide practical solutions where the
domain experts can be more effectively involved. However,
these types of research and their associated publications may
not directly mention DSL keywords. To mitigate this threat
-
 noticed that rarely
the term DSL is left off completely.
Other two possible threats to the validity of our study are:
Search engines providing incoherent information in BibTeX
and, to mitigate this threat, we developed a tool to extract
BibTeX information which considers the peculiarities of
each search engine, reducing the number of possible
mistakes; and we may have not selected the most
representative studies but, to mitigate this threat, we revised
the paper selection spreadsheet in pairs until we reached a
common sense.
VI. CONCLUDING REMARKS
The main motivation for this work was to investigate the
state-of-the-art in engineering DSLs, through systematically
mapping the literature in order to determine what issues have
been studied, as well as by what means, and provide a guide
to aid researchers in planning future research.
After performing this mapping study, we catalogued
1440 relevant studies from an initial set of 4450, which
helped us to investigate several approaches regarding
different aspects of DSL engineering. Our findings could
show which are the domains where DSLs are most suitable.
For instance, four domains of applications draw our
attention, as following (with the respective number of
publications): Web (141), Network (91), Data Intensive Apps
(81), and Control Systems (85). In addition, we were able to
catalogue which types of DSL are being created
(internal/external DSL, DSML, ADL, DSAL), we listed
several techniques, methods/processes to handle DSL, and
we identified different tools to create and maintain DSLs,
including language workbenches.
Moreover, in Figure 6, this study presents a bubble chart
with a full cross reference view of DSL research types and
their respective domains. This way, it is easy to identify
which areas in this research field have been deeply explored
and which are lacking attention from academy/industry with
only a few publications listed.
In our future agenda, we will investigate more deeply the
area of language workbenches through a SR, gathering even
47 60 19 516 51 60 44 10 32 17 21 22 92 32
2 7 4 4 6 11 10 3 2 3 4 8 10
3 5 2 7 2 5 3 6 2 7 16 16 4
6 5 1 1 3 1 2 8 6 2
6 3 4 6 3 2 1 3 1 3 4 14 10
4 7 4 6 12
1 1
20 11 5 4 18 5827255 26 8
Domains
DSL Research Type VS Top 15 Domains
External DSL
Internal DSL
Techniques
Methods/Processes
Tools
ADL
DSAL
DSML
Control
Systems
Data Intensive
Apps
Dynamic
Systems
Education Embedded
Systems
Low-level
Software
Network Parallel
Computing
Real-time
Systems
Security Simulation Testing Visual
Language
Web Others
185Copyright (c) IARIA, 2012. ISBN: 978-1-61208-230-1
ICSEA 2012 : The Seventh International Conference on Software Engineering Advances

more evidence of the area. Moreover, we intend submit an
extended version of this study to a journal because the page
limit here is restraining us to present more details.
ACKNOWLEDGMENTS
This work was partially supported by the National
Institute of Science and Technology for Software
Engineering (INES [67]), funded by CNPq and FACEPE,
grants 573964/2008-4, APQ-1037-1.03/08 and APQ-1044-
1.03/10 and Brazilian Agency (CNPq processes number
475743/2007-5 and 140060/2008-1).
REFERENCES
[1] M. Fowler, Domain-Specific Languages, 1st ed. Addison-Wesley
Professional, 2010, p. 640.
[2]     
domain-ACM Computing Surveys (CSUR), vol.
37, no. 4, pp. 316344, 2005.
[3]         
ACM SIGPLAN Notices, vol. 13, no. 8, pp. 61
99, 1978.
[4] K. Petersen, R. Feldt, S. Mujtaba, and M. Mattsson 
      12th International
Conference on Evaluation and Assessment in Software Engineering,
2008, pp. 7180.
[5]        
Communications of the ACM, vol. 9, no. 3, pp. 157166, 1965.
[6] Communications
of the ACM, vol. 29, no. 8, pp. 711721, 1986.
[7] -specific languages:
 ACM SIGPLAN Notices, vol. 35, no.
6, pp. 2636, Jun. 2000.
[8]          
Electronic Notes in Theoretical Computer Science, vol. 264, no. 5,
pp. 103121, Jul. 2011.
[9]       literature
       
EBSE Technical Report. EBSE-2007-01, 2007.
[10]       -specific
Software, IEEE, vol. 26, no. 4, pp. 1314, 2009.
[11] T. R. Henr      
ACM SIGCSE Bulletin, vol. 37, no. 1, p. 7, Feb.
2005.
[12]    -specific language for
    Proceedings of the
2nd workshop on Domain specific aspect languages, 2007, p. 4es.
[13]        -specific language for
Proceedings of the 5th
international conference on Generative programming and
component engineering - GPCE ’06, 2006, pp. 285294.
[14]         
Proceedings of the ACM international conference companion on
Object oriented programming systems languages and applications
companion, 2011, pp. 5558.
[15] A. Manjunatha, A. Ranabahu, A. Sheth, and K. Thirunarayan,

     2010 IEEE Second
International Conference on Cloud Computing Technology and
Science, 2010, pp. 496503.
[16] -based kinematics
Proceedings of the
ninth international conference on Generative programming and
component engineering, 2010, pp. 157166.
[17]          
development of domain- Computer Science and
Information Systems, vol. 8, no. 2, pp. 317342, 2011.
[18] P. Moreno-Ger, R. Fuentes-Fernández, J.-L. Sierra-Rodríguez, and
B. Fernández--
Information and Software Technology, vol. 51, no. 3, pp. 564580,
2009.
[19]          -oriented
domain-specific language for mas  Proceedings of the Early
Aspects at ICSE Workshops in AspectOriented Requirements
Engineering and Architecture Design, 2007, pp. 411.
[20] P. Sawyer, N. Bencomo, D. Hughes, P. Grace, H. J. Goldsby, and B.
       namically Adaptive
 Second International Workshop on
Requirements Engineering Visualization REV, 2007, pp. 110.
[21]          -
Specific Language to study the interactions between malaria
     International Conference on
Bioinformatics Computational Biology, 2008, pp. 747752.
[22]    -Specific
Language and IDE Tooling to produce Real World Applications for
   Proceedings of the ACM international
conference companion on Object oriented programming systems
languages and applications companion, 2010, pp. 123128.
[23]        -
Specific Languag       
IEEE Transactions on Software Engineering, vol. 37, no. 4, pp. 544
558, 2011.
[24] 
   Proceedings of the 22nd international
conference on Implementation and application of functional
languages (IFL’10), 2010, pp. 153168.
[25]          
   Proceedings of the 41st
annual conference on Design Automation DAC 04, 2004, pp. 924
927.
[26] M. Jiménez, F. Rosique, P. Sánchez, B. Álvarez, and A. Iborra,
  -    
Software, IEEE, vol. 26, no. 4, pp. 3038, 2009.
[27]      h to Constructing Software
   IEEE Transactions on Software
Engineering, vol. 10, no. 5, pp. 567574, 1984.
[28]         Proceeding of the
24th ACM SIGPLAN conference companion on Object oriented
programming systems languages and applications - OOPSLA ’09,
2009, pp. 949955.
[29]        
understanding domain-    IEEE
Transactions on Knowledge and Data Engineering, vol. 14, no. 1,
pp. 172181, 2002.
[30]       
     Science of Computer
Programming, vol. 68, no. 1, pp. 220, Aug. 2007.
[31] J. Evermann and Y. W  
    IEEE Transactions on Software
Engineering, vol. 31, no. 1, pp. 2137, Jan. 2005.
[32]         
     Proceedings of the
IEEE, vol. 93, no. 2, pp. 387408, Feb. 2005.
[33]          
Fourth International Conference on
Generative Programming and Component Engineering (GPCE),
2005, pp. 2946.
[34] F. Lagarde, H. Espinoza, F. Terrier, C. André, and S. Gérard,
       
  FASE’08/ETAPS'08 Proceedings of the Theory and
186Copyright (c) IARIA, 2012. ISBN: 978-1-61208-230-1
ICSEA 2012 : The Seventh International Conference on Software Engineering Advances

practice of software, 11th international conference on Fundamental
approaches to software engineering, 2008, vol. 4961, pp. 116130.
[35]    

5th IEEE International Conference on Industrial Informatics, 2007,
pp. 885890.
[36]         
Compliant Ways to Dene Domain Speci  
ACM/IEEE 10th International Conference On Model Driven
Engineering Languages And Systems (MoDELS 2007), 2008, pp.
4758.
[37]         
context-free grammars for domain-  
Proceedings of the Eighteenth International Conference on Software
Engineering and Knowledge Engineering - SEKE’06, 2006, pp. 363
368.
[38]          
        10th
International Conference Model Driven Engineering Languages and
Systems (MoDELS 2007), 2007, pp. 286300.
[39]           
defining domain-    Proceedings of
the 10th ACM international conference on Generative programming
and component engineering - GPCE ’11, 2011, pp. 127136.
[40] 
to define a language for modelling controlled flexibility in software
IET Software, vol. 4, no. 6, p. 396, 2010.
[41] -specific language design
   Proc. 10th IEEE Int’l Symp. Object and
Component-Oriented Real-Time Distributed Computing, 2007, pp.
29.
[42]        ICSE ’08
Proceedings of the 30th international conference on Software
engineering, 2008, pp. 425441.
[43]         
Proceedings of the 13th international conference on Software
engineering - ICSE ’08, 2008, pp. 831834.
[44]           
Proceedings of the FSE/SDP workshop on Future of software
engineering research - FoSER ’10, 2010, pp. 389392.
[45]     
     
International Journal on Software Tools for Technology Transfer,
vol. 12, no. 5, pp. 353372, 2010.
[46]         
        
ACM/IEEE 11th International Conference on Model Driven
Engineering Languages and Systems (MoDELS 2008), 2009, pp.
167180.
[47]         
      2007 40th Annual
Hawaii International Conference on System Sciences (HICSS’07),
2007, pp. 258267.
[48]  
with multiple domain-   Proceedings of the
IASTED International Conference Software Engineering SE 2009,
2009, pp. 110.
[49] P. Thiemann, -specific language for type-safe
server-   ACM Transactions on Internet
Technology, vol. 5, no. 1, pp. 146, Feb. 2005.
[50]         
Framework-Specific Modeling Langu IEEE Transactions on
Software Engineering, vol. 35, no. 6, pp. 795824, Nov. 2009.
[51] IEEE Software, vol. 27, no.
2, pp. 56  64, 2010.
[52]          
system for domain-Companion of the
18th annual ACM SIGPLAN conference on Object-oriented
programming, systems, languages, and applications - OOPSLA’03,
2003, pp. 2837.
[53]       -based approach to business
       
ICSOC’06 Proceedings of the 4th international conference on
Service-oriented computing, 2007, pp. 166177.
[54] W. Hummer, P. Gaubatz, M. Strembeck, U. Zdun, and S. Dustdar,
  approach for identity and access management in a
   Proceedings of the 16th ACM symposium on
Access control models and technologies - SACMAT’11, 2011, pp.
2130.
[55] J. Boubeta-Puig, I. Medina-Bulo, and A. García-Domínguez,
       -
2011 IEEE Fourth International
Conference on Software Testing, Verification and Validation
Workshops, 2011, pp. 398407.
[56]  IEEE Software, vol. 24, no. 4, pp.
2021, Jul. 2007.
[57]            
2007 29th International Conference on
Information Technology Interfaces, 2007, pp. 753758.
[58] M. Freuden
Proceedings of the Tenth Workshop on Language Descriptions Tools
and Applications, 2010, pp. 11:111:7.
[59] Communications
of the ACM, vol. 52, no. 11, p. 60, Nov. 2009.
[60]          
implementing domain-   Proceedings. Fifth
International Conference on Software Reuse, 1998, pp. 143153.
[61]       - Implement your Language
    Proceedings of the ACM
international conference companion on Object oriented
programming systems languages and applications companion -
SPLASH’10, 2010, pp. 307309.
[62]   ling made easymetaedit+ (tool
  Generative Programming and Component
Engineering (GPCE 2005), 2005, pp. 442446.
[63]           
ACM SIGPLAN Notices, vol. 45, no. 10, p. 444, Oct. 2010.
[64]       
   Proceedings of the 13th international
conference on Model driven engineering languages and systems
Part II (MoDELS 2010), 2010, pp. 3246.
[65]    modeling tools: overview and comparison of
   Proceedings of the ACM international
conference companion on Object oriented programming systems
languages and applications companion - SPLASH ’10, 2010, pp.
139148.
[66] M. Völter and 
   Proceedings of the ACM international
conference companion on Object oriented programming systems
languages and applications companion - SPLASH ’10, 2010, pp.
301304.
[67] National Institute of Science and Technology for Software
Engineering (INES). Available in: www.ines.org.br. Last accessed in
September, 2012.
187Copyright (c) IARIA, 2012. ISBN: 978-1-61208-230-1
ICSEA 2012 : The Seventh International Conference on Software Engineering Advances