Requirements Elicitation Approaches: A Systematic

Review
Aldrin Jaramillo Franco
Engineering Faculty - University of Antioquia
Medelln Colombia
aldrin.jaramillo@udea.edu.co
Abstract: in recent years, the Requirements Engineering
community has dedicated a lot of efforts in order to tackle the
Requirements Elicitation (RE) problem. Although important
advances have been reached, the RE processes still present
challenges that remain between the most critical research topics in
the Requirements Engineering community agenda. In order to
understand the progress that has been made in the RE field we ask:
what approaches exist which supports RE in software development
processes? To answer this question we have made a systematic
review of works performed during the last 25 years (1989 2014)
resulting in 497 publications. From these results, this paper reports
on the main characteristics of each proposal like: purpose, sources
of requirements required, target produced, type of knowledge
representation used, and types of resources, methods and tools
required to accomplish their goal. We also identify the prominent
issues of interest for the researchers, and the most influential works
and trends over time. We argue that the results of this work are
relevant toward understanding the state of the art in RE, providing
insights on the relevant issues and perspectives that should be
considered in future proposals.
Keywords: Requirements elicitation; Systematic literature review,
State of the art in requirements elicitation.

I.

INTRODUCTION

This paper aims at understanding the progress that has been

made in the RE field during the last 25 years. More specifically
we are interested in answer the research question: what
approaches exist that supports RE in software development
processes? We detail this overarching question by investigating
relevant characteristics of the identified works. To accomplish
this purpose, taking into account the work of Kitchenham et al.
[1], [2], we have conducted a Systematic Literature Review
(SLR), which results in 497 publications. This paper presents
contributions for researchers and practitioners who can have a
better understanding of the RE evolution during the last 25
years. In addition, the obtained results provide insights on the
relevant issues and perspectives that should be considered in
future proposals. The resulting repository of RE approaches is
available online; we hope that this resource will facilitate the
development of future works aimed to contribute to the RE field.
II.

SYSTEMATIC LITERATURE REVIEW METHODOLOGY

In order to answer the overarching question RQ0. what

approaches exist which support requirements elicitation in
software development processes? we have defined a set of

inclusion / exclusion criteria, which delimit the scope of the

Systematic Literature Review (SLR); a summary of these
criteria is presented in Table I.1 The search was limited to the
last 25 years (since 1989 to 2014). Some definitions used to
establish the study scope are: Approach: an approach is a
systematic arrangement, usually in steps, of ideas or actions
intended to deal with a problem or situation [3]. Method: a
method consists of heuristics and guidelines for the
requirements engineer at different stages of a process. [4].
Technique: a technique is a way of doing something or a
practical method applied to some particular task [3].
For each identified approach, we are interested in solving the
following detailed research questions (Table II): RQ1. What
sources of requirements are required by the approach? RQ2.
What is the purpose of the approach? What targets are produced
by the approach? RQ3. What knowledge is represented in the
approach? How is represented? RQ4. How the requirements are
discovered? What methods are used by the approach? In our
SLR we followed a two-steps process; the first step made a
systematic search in several scientific databases: IEEE,
SPRINGER, ACM, DBLP and SCOPUS. The second step
consisted in a snowball process over the publications obtained
from the previous step. Fig. 1 summarizes the SLR process.
Considering our research questions and scope we defined the
following search string: (elicitation OR gathering OR
acquisition OR discovery) AND (requirements OR
functional requirements OR non-functional requirements)
AND (approach OR proposal OR method). We
consolidated the results in a document which is available
online. 2 This repository of tagged proposals facilitated the
answer to the research questions as presented in the next section.
Note: some of the selected approaches make reference to
Requirements Engineering approaches; nevertheless, they
describe in detail their Requirements Elicitation process due this
we have considered relevant for our research.
III.

In relation to RQ0. What approaches exist which allow the

requirements elicitation in software development processes?
The identified publications were found in the following
1
2

978-1-4673-6630-4/15/$31.00 2015 IEEE

SLR RESULTS

Tables and graphics can be found in http://goo.gl/OPcHmp

http://goo.gl/b2wI5p

scientific databases: IEEE provided 56% (279 publications),

SPRINGER 23% (114 p.), ACM 9% (46 p.), SCOPUS 6% (29
p.) and other databases 6% (29 p.). Fig. 2 shows the publications
per database. We noted that most of publications (56%) were
found in IEEE database being preferred by the RE community to
date. Fig. 3 shows that 68% (340 p.) corresponds to conferences
followed by journals 27% (133 p.); 4% (19 p.) are book chapters
and 1% (5 p.) corresponds to books. These results show how
conferences are by far the main mean of knowledge socialization
for the RE community. In order to establish the evolution of the
RE field in the last 25 years, we have defined two periods of
comparison: period 1, since 1989 to 2001 and period 2, since
2002 to 2014. Fig. 4 presents the number of publications in each
of these periods. We observe a remarkable increase of proposals
(362 p.) during the last period in comparison with the first one
(135 p.); this puts in evidence the progressive interest and effort
dedicated by researchers and practitioners in the search for
solutions to the challenges posed by RE processes. This also
could be an important indicator of the growing relevance of RE
issues not only for academia but also for industry. RQ1.
Sources of requirements. Table III shows that the main sources
in both periods are: Domain knowledge, Initial requirements and
Stakeholders goals. We can see that in period 1, researchers
consider that these sources are equally relevant; nevertheless,
this changes in period 2 where Stakeholders goals are given
priority (63 p.) over Initial requirements (39 p.) and Domain
knowledge (37 p.). We can also observe how some important
sources in period 1 (ERP and business processes, Situations,
arguments, selected strategies and options, and Use cases) have
lost relevance in the second period. Likewise, in the last period
we note the emergence of sources like Business process models,
Security goals, privacy goals and attacks, and Legal texts. RQ2.
Purpose and target. Fig. 5. The first period shows a dominant
line of works in FRs (60 p.); meanwhile, proposals on NFRs
appear in an incipient form (7 p.). The elicitation of FRs and
NFRs also appears in an important manner being the second
research stream (46 p.). Other works (i. e. groundwork)
constitute the third priority for researchers (22 p.). On the other
hand, important changes occurred in the second period in
comparison to the first one: the main line of research
corresponds to FRs and NFRs (143 p.), the second priority is the
line of FRs (118 p.), in third place we find research on NFRs (70
p.) and the fourth line of work corresponds to other proposals
(31 p.). RQ3. Knowledge and representation used. The main
forms of knowledge representation used in both periods are
Scenarios and Goal models. In the first period researchers use
Scenarios and Goal models in similar proportions (21 p. and 19
p. respectively). Otherwise, in the last period there is a wide
preference for Goal models (75 p.) over Scenarios (32 p.). We
also observe that Viewpoint models, Conceptual meta-models
and Goal obstacles are diminishing in their use during the last
period. Besides, we note the emergence of Security models, Use
cases and Ontologies. These results are condensed in Table IV.
RQ4. Methods. Considering methods, we have grouped the
approaches into categories which extend those proposed by van
Lamsweerde [5] and Wieringa and Daneva [6]. Table V
summarizes these results. Most used categories in both periods
are Reference-model-based approaches, Goal-based reasoning
and Scenario-based elicitation and validation. Reference-modelbased approaches are the most used methods in both periods.
However, we observe in the first period that Scenario-based
elicitation and validation approaches (31 p.) are more used than
Goal-based reasoning approaches (21 p.); this situation changes
drastically in the second period where Goal-based reasoning
approaches (75 p.) are preferred over Scenario-based elicitation
and validation approaches (38 p.). In contrast with the first

period, in the last years we perceive an increasing interest in

methods like Creativity and collaborative-based approaches,
Pattern-based approaches, Quality-model-based approaches,
Quality-verification-based approaches and Requirements reuse
among others. Finally, in order to identify the most influential
proposals, we have consulted and consolidated the number of
citations of each publication in IEEE, ACM, SCOPUS and
Google scholar databases; the main of these results appear in
Table VIII. Considering these outcomes, it is important to
highlight the influence of the works of van Lamsweerde et al.
who has 4 publications in the top ten most cited papers;
Mylopoulos et al. 3 p., Scheer 1 p., Yu 1 p. and Rolland et al. 1p.
We also note that these publications are mainly related to
Agents-based approaches (REA-458, 265, 268), Referencemodel-based approaches (REA-417), Groundwork (REA-452),
Goal-based reasoning (REA-492, 454), Quality-model-based
approaches (REA-453, 274) and Scenario-based approaches
(REA-357). This puts in evidence the noteworthy interest of the
RE community for these types of solutions.
IV.

CONCLUSIONS

In relation to the Sources of requirements used by the

proposals, we found that in last period, Stakeholders goals are
the most used input. Furthermore, we observe the emergence of
Business process models, Security goals, privacy goals, and
attacks and Legal texts as relevant entries for the approaches.
Regarding the Purpose and target of the approaches, in recent
years (second period) we observe an important increase of
proposals in all lines of research: FRs, NFRs, FRs and NFRs,
and Groundwork. We also note a growing interest especially in
the lines of FRs and NFRs and NFRs elicitation; this could be
an important indicator of the growing relevance of RE issues not
only for academia but also for industry. As to the Knowledge
and representation used, we found that Goal models are by far
the preferred form of knowledge representation in last years,
followed by Scenarios, Security goals, Use cases and
Ontologies. Regarding the Methods used by the proposals, in
recent years we observe a significant growth on the use of Goalbased reasoning approaches; meanwhile, Scenario-based
approaches have lost terrain but still are protagonists in the
scene. We also note a remarkable bet for other methods like
Security and privacy-based approaches, Creativity and
Collaborative-based approaches, Pattern-based approaches,
Quality-model-based approaches, Quality-verification-based
approaches and Requirements reuse among others. We estimate
that these methods have been mainly fostered by challenges of
trends like internet, mobility and the need for innovation among
others; therefore, they will play an important role in the future.
REFERENCES
[1] B. Kitchenham, T. Dyba, and M. Jorgensen, Evidence-based software
engineering, in Proc. of the 26th International Conference on Software
Engineering (ICSE04). pp. 273281. 2004.
[2] B. Kitchenham, R. Pretorius, D. Budgen, O. P. Brereton, M. Turner, M.
Niazi, and S. Linkman, Systematic literature reviews in software engineering
a tertiary study, Information and Software Technology, vol. 52, no. 8, pp. 792
805, 2010.
[3] D. Zowghi and C. Coulin. Requirements Elicitation: A Survey of
Techniques, Approaches, and Tools. In: Aurum, A. and Wohlin, C. (eds.),
Engineering and Managing Softaware Requirements. pp. 19 46. Springer Berlin
Heidelberg. 2005.
[4] B. Nuseibe and S. Easterbrook. "Requirements Engineering: A Roadmap".
Proceedings of the Conference the Future of Sofware Engineering. ICSE 00.
Limerick Ireland 2000.
[5] A. van Lamsweerde. Requirement Engineering in the Year 00: A Research
Perspective. Proc. 22nd Int. Conf. on the Software Engineering, University of
Limerick, Ireland. 2000.
[6] R. Wieringa and M. Daneva. Requirements Engineering for Enterprise
Systems:WhatWe Know and What We Dont Know?. S. Nurcan et al. (eds.),
Intentional Perspectives on Information Systems Engineering, 2010.