A Business-Oriented Approach to Requirements Elicitation
Adam Przybyłek
Faculty of Mathematics, Informatics and Mechanics, University of Warsaw, Warszawa, Poland
Keywords: Requirements, Business Process Modeling, BPM, Use Cases, Traceability, UCDA, UML, Activity Diagram.
Abstract: “The hardest single part of building a software system is deciding precisely what to build” (Brooks, 1987).
Faulty requirements analysis is a major reason for project failure or unsatisfactory information systems that
do not fulfill business needs. Although it has been long recognized that system requirements can only be
understood within the context of the business processes and business modeling has become the initial phase
of most software processes, the transition between business modeling and requirements gathering is still a
challenge for research. Moreover, existing work in this area tends to introduce accidental difficulties. This
paper reports the results of an action research conducted for elaborating a Business-Oriented approach to
Requirements Elicitation. Our approach integrates Requirements Engineering with Business Process
Engineering and derives system requirements based on business process models. This ensures that system
requirements meet real business needs. The proposed approach is illustrated by a real-world example.
1 INTRODUCTION
Information systems (ISs) sustain the underlying
business processes in most of today’s organizations.
However, few ISs are developed with explicit
consideration of the business processes they are
supposed to support. A recent survey (Wolf and
Harmon, 2012) shows that in 179 out of 372
organizations that were examined, the support
provided by the existing software systems was never
or occasionally consistent with the business
processes. The sources of this misalignment can be
traced to the “essential” and “accidental” difficulties
of the requirements engineering process. We
organized these difficulties in Table 1. For each
essential difficulty we identified the corresponding
accidental difficulty that compounds it. We also
pointed out the challenges that, if met, would reduce
the accidental problems. The essential difficulty is
inherent in what one is trying to accomplish; it is
part of the problem itself. In turn, the accidental
difficulty is introduced through inadequate practices;
it comes from the way one is tackling a problem
(Brooks, 1987). Thus, analysts should focus their
attention on solutions for the essential difficulties,
and apply techniques and methods which help them
avoid the accidental difficulties.
We explain how to read Table 1 by taking the
first row as an example. Understanding what the
customer needs is an essential difficulty, because
customers do not really know what they need.
System analysts have to make the customer realize
its needs. Thus, this difficulty is inherent in the
problem of requirements elicitation and we cannot
avoid it. However, inadequate requirements
elicitation practices unnecessarily exacerbates this
difficulty. For instance, customers are not involved
enough in the requirements engineering process or
system analysts just model what the customer is
saying, while customers usually can not articulate
requirements that truly reflect their business needs.
The challenge is to build relevant business process
models and then to derive system requirements from
these models.
To address the challenges presented in Table 1,
we combine techniques from both the fields of
Business Process Engineering and Requirements
Engineering and define BORE: a Business-Oriented
approach to Requirements Elicitation. Our approach
allows for deriving system requirements from
business process models and enables traceability
between business processes and the corresponding
system requirements. This ensures that system
requirements meet real business needs and that there
are no superfluous requirements.
The rest of the paper is organized as follows.
Next section reports the design of our research.
Section 3 presents the general description of BORE,
while its succeeding subsections describe the stages
of the approach in detail. Section 4 demonstrates
152
Przybyłek A..
A Business-Oriented Approach to Requirements Elicitation.
DOI: 10.5220/0004887701520163
In Proceedings of the 9th International Conference on Evaluation of Novel Approaches to Software Engineering (ENASE-2014), pages 152-163
ISBN: 978-989-758-030-7
Copyright
c
 2014 SCITEPRESS (Science and Technology Publications, Lda.)

Table 1: Essential and accidental difficulties of the requirements engineering process.
Essential difficulties Accidental difficulties Challenges
understanding what the customer
needs
<= customers do not really know
what they need (Faulk, 1997) or have
only a vague picture of their
requirements (Maciaszek, 2005)
inadequate requirements elicitation practices
<= customers are not involved enough in the requirements
engineering process;
<= customers usually can not articulate requirements that
truly reflect their business needs
business analysis must
precede requirements
elicitation in order to obtain a
deep enough understanding
about the organization;
business process models must
drive requirements elicitation
effective communication among
stakeholders due to the “culture gap”
(Taylor-Cummings, 1998) between
business and system domains;
<= a requirements specification has
many audiences with different
viewpoints and knowledge (Faulk,
1997)
rework and miscommunications due to many notations
<= business processes are not consistently documented and
different stakeholders tend to use different notations,
conventions and techniques to represent them (Monsalve et
al., 2010);
<= business models are usually done using different
notation than that of software models; a recent survey
(Harmon & Wolf, 2011) shows that the vast majority of
business people (72%) use BPMN, while UML, which is a
standard in software design, was listed only by 18%
respondents (the respondents could indicate more than one
notation; 8% chose EPC).
if we could use the same
notation through the whole
project all stakeholders can
share the same work products
frequent and arbitrary changes to
requirements (Faulk, 1997)
<= business processes change due to
technology advancements, pressure
from competitors, new legislation,
etc.; in this case, the requirements
may need adaptations to align the
evolving business processes
deficiencies in backward traceability (Ravichandar et al.,
2007; Gotel & Finkelstein, 1994) make it difficult to keep
consistency between documentation and the underlying
system
<= business processes are not linked to the system
requirements and thus evolve independent from the IS;
the documentation must link
business processes to
artefacts of an analysis,
design and implementation in
an explicit and traceable
manner;
how to use BORE. Section 5 discusses related work.
Finally, Section 6 concludes with a review of the
contribution.
2 RESEARCH DESIGN
This research reports on the action research project
at a private university. The university was
experiencing problems with an overwhelming
amount of work during the admission and
timetabling period. It was usual for administrative
staff to work overtime, evenings or weekends.
Moreover, the highly competitive market of
educational services was forcing the university to
improve the teaching process and communication
among lecturers and students by leveraging IT
solutions. Nevertheless, the authorities did not have
a clear and detailed idea about the real needs and
were not completely sure what they wanted.
The action research methodology aims at
providing value to the participant organization but
acquiring new knowledge at the same time (Sjoberg
et al., 2007). The research objective was to develop
a new approach that (1) supports requirements
elicitation under conditions of uncertainty about
client needs; and (2) makes that the system
requirements are in alignment with and provide
support for the underlying business processes. The
practical objective was to analyze the business
situation, identify opportunities for improvements,
and specify the requirements of an IS that supports
the business needs.
The participant organization was selected for this
research due to: (1) its willingness to undergo a
process of innovation; (2) its interest for the project
that was founded from an external grant (it was free
for the organization); (3) its accessibility, and (4) the
application domain that was known by the
researcher.
Four main approaches to data collection were
used in this research. These were: semi-structured
and unstructured interviews, apprenticing,
workshops, and document analysis.
The researcher met with seven customer
representatives at every level in the organization.
These representatives were selected by the research
coordinator, who was an employee of the university
and knew the staff. Then, the representatives were
ABusiness-OrientedApproachtoRequirementsElicitation
153

trained in business process diagrams, activity
diagrams and use case diagrams. As part of their
training, they had to model business processes and
draw a use case diagram based on a given case
study.
The customer stakeholders participated in the
project to varying degrees. During As-Is business
process modeling, the researcher created models and
then the customer stakeholders were consulted to
validate the models. This approach is called
consultative participation. In contrast during
business process engineering and use-case modeling,
the customer stakeholders collaboratively developed
models in facilitated group sessions. This type of
participation is consensus-driven in the sense that it
is the stakeholders who “own” the model and hence
decide its contents (Bubenko et al., 2001), while the
main role of the researcher is to facilitate the
discussion, make suggestions, and gain feedbacks.
Various notations have been proposed to model
business processes, e.g.: BPMN (Business Process
Model and Notation), UML, Petri Nets, DFD (Data
Flow Diagram), IDEF3 (Integrated DEFinition
Method 3), RAD (Role Activity Diagrams) and EPC
(Event Driven Process Chain). Today, the most
prominent among these are BPMN and UML
Activity Diagram. BPMN is more complex but at the
same time more powerful (BPMN has more than
100 graphical elements). Nevertheless,
Marcinkowski (2010) found that activity diagrams
provide adequate modeling support for the purpose
of business analysis.
Since our approach promises seamless transition
from business process models to use case models, it
needs the same notation for both domains. Thus, our
natural choice was to use UML Activity Diagram
and its extension called Business Process Diagram
(if the reader is not familiar with these diagrams, we
recommend reading Eriksson and Penker (2000) and
Wrycza et al., (2005)).
Business Process Diagram introduces a set of
stereotypes that illustrates the interaction between
the processes, the resources that participate in the
processes and the people acting in the processes. It
describes how work is done within the business
environment. The core business modelling element
is represented by stereotyping an activity to a
«process» or by a special icon. A process can span
over multiple swimlanes to represent the fact that
multiple roles can cooperate to its execution.
Resource instances are represented as objects. Input
objects are usually placed on the left of the process
and output objects are placed on the right. A result
produced by one process can constitute input to
another process. The people resource is placed
below or above the process and is linked to it by a
dependency stereotyped by «control». For more
details about Business Process Diagram, the reader
is referred to (Eriksson and Penker, 2000).
3 APPROACH DESCRIPTION
BORE is based on the following observations: (1)
Business Process Improvement (BPI) and IT
planning are closely related; (2) customer
stakeholders are often not completely sure about
their real requirements and are not aware of what
support they can expect from an IS; and (3) business
analysis and requirements elicitation consist of
overlapping activities and bear much resemblance to
each other. These observations give rise to the
corresponding assumptions: (1) BPI should be
viewed as preliminary to and integrated with the
development of IS; (2) understanding the business
processes is a precondition for adequate
requirements elicitation; and (3) business process
models can be used for requirements elicitation.
Wherever possible, we attempted to leverage
existing, well-established techniques, methods and
notations in business analysis and requirements
engineering rather than developing new ones.
Particularly, we borrowed such means as Joint
Application Design, interviewing, apprenticing,
meta-plan, use-cases, and Eriksson-Penker Business
Extensions.
BORE is a three-stage approach (see Figure 1).
These stages are presented in subsequent
subsections.
As-Is Business
Process Modeling
Business Process
Improv ement
Functional
Requirements
Elicitation
As-Is
models
To-Be models
[draft]
list of processes to be
computer-aided
To-B e m o d e l s
use-case models
traceability matrix
Figure 1: Overview of BORE.
ENASE2014-9thInternationalConferenceonEvaluationofNovelSoftwareApproachestoSoftwareEngineering
154

3.1 As-Is Business Process Modeling
The first stage of our approach aims at
understanding the organization for which an IS is
going to be developed. Business process models are
the main deliverables of this stage. Several
techniques are available to transfer knowledge from
customer stakeholders to analysts. At this stage, we
recommend interviewing, apprenticing and looking
through existing organizational documents.
Interviews can be guided by focus questions such
as: What are the main processes of the
organization?; How are these processes related?;
What activities do these processes consist of?;
Which information and material flows do these
processes consume and produce?; Who is
responsible for performing and supporting these
processes? (Bubenko et al., 2001). A complementary
way to obtain this information is through mining
event logs (if ones exist) generated by legacy
systems (Przybyłek, 2013a; 2013b; 2013c).
The gathered information is used to outline an
overall picture of the business on process diagram
(Eriksson and Penker, 2000). The diagram must
include all fundamental enterprise-wide processes.
Each complex process contains a number of
activities that are performed as part of the process.
Each of these activities can be considered a process
of its own and as a sub-process to the containing
process. Thus, we can progressively apply functional
decomposition and model the internal structure of
each complex process at a lower level activity
diagram (Przybyłek, 2007). A process can be broken
down into smaller sub-processes as long as an
elementary level is achieved. A process is
elementary if it is performed in one location at one
time and leaves the business in a consistent state
(Frost and Allen, 1996). The obtained diagrams
should be validated by customer stakeholders.
Usually, several iterations are needed to get the final
versions.
Probably the most efficient technique for
understanding the details of business processes and
their connections is apprenticing. In this technique,
the analyst is the apprentice, with the worker as the
master craftsman. The analyst sits alongside the
worker at the normal workplace and receives a
running commentary on the work as it happens.
Almost everyone is good at explaining what they are
doing while they are doing it. As the work is
observed and explained, the analyst sketches a
model of each task and its connections with the other
tasks. As the models are built, the analyst feeds them
back to the worker to obtain confirmation that they
are correct and to raise questions about any areas of
uncertainty (Robertson, 2013).
The resulted models (As-Is) describe how the
business is working today at any level of abstraction
from enterprise-wide processes to single tasks
performed by single people. For various guidelines
that can support business process modeling, the
reader is referred to (Vara, 2011).
3.2 Business Process Improvement
Once the business processes have been modeled, we
have to agree what part of the business is to be
automated. Two important factors that must be
considered are costs and benefits. Note, that
automating processes for the sake of automation
does not lead to significant improvements
(Weerakkody and Currie, 2003). Thus, as suggested
by Hammer (1990), instead of blindly automating
manual processes, the processes are reengineered
while taking advantages of the possibilities for
automation.
At this stage a workshop with key stakeholders
should be held. The role of the workshop is to bring
stakeholders together for a common purpose, for a
short, intensive period. The role of customer
stakeholders is to give feedback about the findings
and suggestions presented by the analysts, make
their own proposals, and provide more knowledge
about the current situation (Lehtola et al., 2009). The
most crucial technique at the workshop is
brainstorming, which is one way of inventing.
Brainstorming contains two phases: the generation
phase, where ideas are collected; and the evaluation
phase, where the collected ideas are discussed
(Leffingwell and Widrig, 1999). To foster a creative
atmosphere, all ideas are acceptable, and no one can
slow the process down by criticizing or debating
ideas. The aim is to be as imaginative as possible,
and to get input from all stakeholders, often by using
the ideas of others to trigger a different idea of their
own (Robertson, 2013). Some of the ideas may seem
silly, but they must still be produced to help come up
with the good ideas. An idea may be impracticable
as is, but a mutation of it need not be. When the
idea-generation phase terminates, it is time to initiate
idea reduction (Leffingwell and Widrig, 1999). The
facilitator walks through each idea and asks the
submitter to provide an explanation. Then, the group
discusses the ideas and eliminates those that are not
worthy of further consideration.
The first step at the workshop is to identify
improvement opportunities in the business
processes. By analyzing the existing processes, the
ABusiness-OrientedApproachtoRequirementsElicitation
155

participants look for redundancies and bottlenecks.
This step should be supported by dedicated
simulation software (Gawin and Marcinkowski,
2013). Next, the participants identify and discuss
new organizational needs that could be fulfilled by
IT. They also analyze how IT can reshape the way
business is done today in order to create an
environment in which the staff exploits information
more effectively. Since new business processes may
occur after this stage, the process models generated
at the previous stage must be adjusted.
Once elementary business processes (EBPs) are
revised, they must undergo a prioritization process
to decide which of them and in what extent we are
going to computerize. For each EBP there are three
possibilities. A process may be either performed
automatically by an IS, or supported by an IS, or
performed manually. The abbreviations “A”, “S”,
and “M” in the process descriptions represent these
situations. A process is automated when the IS
performs it without human participation. A process
is supported when a user interacts with the IS to
perform the process.
A survey of prioritization techniques can be
found in Bendakir et al., (2013). In our approach, for
each EBP customer stakeholders consider how an IS
may help a worker complete the process and assess
the potential benefit using a Low-Medium-High
scale. In turn, supplier stakeholders estimate the cost
of computerization. In many cases, one EBP can be
computer-aided more readily as a result of having
computerized another EBP. If computerization of a
certain EPB is irrational without computerization of
other EPBs the cost is estimated with an assumption
that the closely dependent EPBs are also
computerized. All closely dependent EPBs are
explicitly listed for each process.
3.3 Functional Requirements
Elicitation
At this stage, workshop is still the main technique
for gathering information and collaborative decision-
making. The business processes that have been
designated to computerization, are used to derive
functional requirements that the IS should supply
(and that the business needs). When reviewing
literature, several concrete procedures on how to
transform business process models into use case
models can be found (Dijkman and Joosten, 2002);
(Štolfa and Vondrák, 2004); (Vara et al., 2008).
However, from our experience, it has seemed to us
that the transformation process cannot be
algorithmized, and so we provide only general
guidelines that support this stage. First of all, we
must look from both the perspective of the system
(asking what will make up a well-defined use case?)
and from the perspective of the business process
(asking what is needed from the IS?) (Eriksson and
Penker, 2000). A well-defined use case must specify
a complete functionality which yields an observable
result of value for one or more actors.
We suggest to start the transformation from the
processes labeled with ‘S’. Usually, we must create
a use case for each such process. Then, the partition
at which this process is occurring is mapped to an
actor associated with the use case. Next, we move to
the processes labeled with ‘A’. Usually, they are
computerized as a part of the other use cases that
have been already created. If we mapped an entirely
automated process to its own use case, the use case
would not be associated with any actor. According
to UML, this situation is not valid.
Note, that an organization may change due to the
IS deployment, and the change may have an effect
on the business processes. Usually, some EBPs
disappear, and the new ones that reflect new ways of
running the business are introduced. Thus, we must
once again adjust the process models. The new
models (To-Be) are designed from the original ones
(As-Is) and the support that the IS provides (Vara et
al., 2008).
It is important to ensure that the IS will solve the
real business needs. Therefore each use case
proposed for the system must have its origin in at
least one business process (a use case is said to be
traced from that/this process). Note that, it is
possible to have several use cases coming from the
same business process as well as a single use case
supporting more than one process. Among numerous
techniques for tracing use cases (summarized in
(Gotel and Finkelstein, 1994) and (Torkar et al.,
2012)), we recommend traceability matrices. A
traceability matrix is typically implemented as a
table or a spreadsheet. The processes are associated
to the rows and the use cases are associated to the
columns of a matrix. When a process is related to a
use case, a mark is placed in the intersecting cell.
When the use case diagram and the traceability
matrix are accomplished, the next step is to
document every identified use case in detail. The
additional information is discovered by interviewing
potential users of the system. Then, all use cases are
analyzed to solve conflicts and inconsistencies. Here
techniques and approaches from traditional
requirements analysis (see (Maciaszek, 2005)) may
be applied. The resulted use case model specifies a
software system that adequately supports the
ENASE2014-9thInternationalConferenceonEvaluationofNovelSoftwareApproachestoSoftwareEngineering
156

business processes of the organization and is the
starting point for the rest of the development phases.
4 BORE IN ACTION
This section reports a project conducted in a private
university in Poland. The university staff provided a
set of narratives of the current business processes.
The employee who was most knowledgeable about
the issue of interest was interviewed. For the sake of
space and brevity, we omit some of the processes,
simplify others, and present only a few artifacts
developed during the project. Figure 2 gives a top
level overview of the main business processes, while
a brief description of each process is reported below.
Admission Process (Figure 3). As a formal
prerequisite to be eligible for admission to
Bachelor's program, an applicant need to have a
matriculation certificate. Admission to Master's
program requires the satisfactory completion of an
undergraduate (bachelor's) degree. All candidates
must submit an application form and a copy of the
relevant diploma. They must also pay a non-
refundable recruitment fee. Admission Committee
ensure each application is complete. If the number of
applicants exceeds the number of available places,
the applicants undergo a process of evaluation.
Candidates for undergraduate level are evaluated on
the basis of their matriculation grades. Master's
program require candidates to pass an entrance
examination. The exam is prepared, conducted and
reviewed by Examination Board. Next, Admission
Committee draws up a list of the strongest
candidates who are initially accepted. Before the
final list of the admitted students is announced,
candidates must pay all tuition fees for the academic
semester and send their original diplomas.
Analysis of the Business Environment. The
chancellor and heads of departments continuously
track changes in job offers, to adjust the curriculum
to the needs of the market. IT magazines, job portals
and websites of other universities are main sources
of the knowledge. The authorities also have to take
into account program basics defined by the
government.
Marketing. The approved curriculum must be
presented to a wide range of potential students.
Thus, the PR staff create marketing materials that
are disseminated through press, radio and TV. They
also elaborate a content of the university's website.
Evaluation of Lecturers. After each semester an
assistant to the dean ranks lecturers on the basis of
surveys and visitations. The surveys are conducted
among students by the assistant, while the visitations
are carried out by heads of departments.
Assigning Courses to Lecturers. Heads of
departments have to map courses to lecturers. First,
courses are offered to those lecturers who were
employed in the previous academic year and who
were highly rated by students. If a new lecturer is
employed, his/her research background is evaluated
and the payment is negotiated individually. The final
decision about the employment is taken by the
chancellor. The assignment of a lecturer to courses is
set out in a document known as a workload card,
which must be signed by the chancellor.
analysis of the
business
environment
curriculum
assigning
courses to
lecturers
ev aluation of
lecturers
lecturers
ranking
workload
cards
elaborating
courses
syllabus [new]
timetabling
timetable
marketing
adv ertising materials
conducting
classes
«people»
skilled
students
admission
process
list of first year
students
application
forms
w eb content
«people»
unskilled
students
salary
list
syllabus [old]
IT magazines
j ob portals
univ ersity
rankings
w ebsites of other
universities
available
classrooms
lecturers'
preferences
evidence of
payment
Heads of
departments
Chancellor
HR manager
Admissions staff
Dean
Students
Assistant Dean
Lecturers
Planner
PR staff
«control»
«control»
«control»
«control»
«control»
«control»
«control»
«control»
«control»
«control»
«control»
«control»
«control»
«control»
Figure 2: Top level As-Is business process model.
ABusiness-OrientedApproachtoRequirementsElicitation
157

Examination board Admission staff«external» Candidates
enrolling
candidates
paying a
recruitment fee
rev iew ing
exam
drawing up a list of
initially accepted
paying all tuition fees
drawing up a list of
first year stude nts
informing about
exam
v alidating
applications
calculating the
matriculation score
sending original
diplomas
submitting
documents
preparing
exam
conducting
exam
sending acceptance /
rej ection letter
atte nding
exam
[available
places >=
number of
candidates]
[else]
[Master's
program]
[Bachelor's program]
Figure 3: As-Is business model for the admission process.
Table 2: Computerization decision for the admission process.
id elementary business process
benefits from
computerization
cost to implement
dependent
processes
decision
1 submitting documents high low 4 S
2 paying a recruitment fee low high M
3 validating applications moderate high 1 S
4 enrolling candidates moderate low A
5 informing about exam low low 3 S
6 preparing exam low moderate S
7 conducting exam high low 6 S
8 reviewing exam high low 6, 7 A
9 calculating the matriculation score high low 3 A
10 drawing up a list of initially accepted low low 3, 8, 9 S
11 sending acceptance / rejection letter low low 10 A
12 paying all tuition fees low high M
13 sending original diplomas low high M
14 drawing up a list of first year students low low 10 S
Timetabling. On the basis of the workload cards,
lecturers’ preferences and classroom availability, the
timetable is drawn up by a planner.
Elaborating Courses. Once courses are assigned,
lecturers have to prepare corresponding syllabuses
and modules. A lecturer can follow the syllabus that
was used during the previous year or custom it to
his/her own vision. All syllabuses must be approved
by the dean. Detailed course curricula, rules of
obtaining credits, and lists of compulsory literature
are displayed in the showcases.
Conducting Classes. Lecturer familiarize students
with details of curricula, rules of obtaining credits,
and consultation hours. Lecturer conduct classes
according to the timetable.
During the workshop, stakeholders decided to
computerize the following processes: admission,
assigning courses to lecturers, timetabling, and
evaluation of lecturers.
Every single business process was reviewed to
consider the potential support from IT. The results
for the admission process are summarized in Table
2.
Figure 4 shows the To-Be business model for the
admission process, which describes the way in
which the university wants to operate after
deploying the IS. Note, that the anticipated IS is
represented by a new partition, and that the
university requires candidates to apply online.
Instead of labeling each process with “A”, “S”, or
ENASE2014-9thInternationalConferenceonEvaluationofNovelSoftwareApproachestoSoftwareEngineering
158

Legend
paying all tuition fees
sending original
diplomas
SystemExamination board Admission committee«external» Candidates
enrolling in
the system
paying a
recruitment fee
rev iew ing
exam
drawing up a list of
initially accepted
draw ing up a list of
first year students
informing about
exam
v alidating
applications
submitting
documents
calculating the
matriculation score
conducting
exam
preparing
exam
sending acceptance /
rejection letter
performed
manually
supported by a
software system
performed automatically
by a software system
attending
exam
[available
places >=
number of
candidates]
[else]
[Bachelor's
program]
[Master's
program]
Figure 4: To-Be business model for the admission process.
Courses&Lecturers
Admis sion
Candidate
enroll
withdraw
application
take exam
view status
Examination
board
create exam
activate exam
rev iew
applications
adjust admission
settings
create student
list
Admis sion
committee
evaluate
lecturers
«CRUD»
manage
courses
Planner
«CRUD»
manage rooms
«CRUD»
manage
constraints
generate plan
customize plan
HR manager
«CRUD»
manage
lecturers
assign courses to
lecturers
Head of
department
Student
Figure 5: Use case diagram.
“M”, we express this information using
background color.
After accomplishing the To-Be business models,
we applied the guidelines presented in Section 3.3 to
derive use cases (Fig. 5). Next, we traced back the
resulted use cases to their source. Table 3 lists all the
EBPs consisting on the admission process down the
left column. In the row across the top, it lists the use
cases derived to support the stated processes. A
traceability relationship is indicated with an X in the
cell to record the fact that a specific use case has
been defined for the purpose of supporting one or
more business process. Note that a single process
may be supported by multiple use cases.
ABusiness-OrientedApproachtoRequirementsElicitation
159

Table 3: Traceability matrix for the admission sub-processes to use cases.
enroll withdraw application take exam view status review applications create exam activate exam adjust admission settings create student list
enrolling in the system x x x
submitting documents x
validating applications x
informing about exam x
preparing exam x
conducting exam x
reviewing exam x
calculating the matriculation score x x
drawing up a list of initially accepted x
sending acceptance / rejection letter x
drawing up a list of first year students x
paying a recruitment fee
paying all tuition fees
sending original diplomas
5 RELATED WORK
The most recognized technique that enhances the
quality of the deliverables resulting from the
requirements engineering process is the Use Case
Driven Approach (UCDA) introduced by Jacobson
(1987). Being use-case driven means that use cases
bind together all software models. The basic
concepts of UCDA are actors and use cases
(Jacobson, 1992). An actor is a specific role played
by a system user, and represents a category of users
that demonstrate similar behaviour when using the
system. The way an actor uses the system is
described by use cases. A use case specifies a
sequence of actions, including variants, which actor
and system perform in order to yield observable
results of value to a particular actor (Jacobson et al.,
1999). UCDA helps to manage complexity, as it
focuses on one specific aspect of usage at a time. It
also provides means for customers and users to
actively participate in requirements analysis, as use
cases are expressed in terms familiar to them
(Regnell et al., 1996). However, UCDA does not
present any concrete heuristics to guide the
requirements elicitation process and does not
provide explicit means which could be used for
linking business processes and use cases. These
lacks may result in use cases without any underlying
rationale. BORE improves the original UCDA by
integrating it with business process engineering and
by providing an alternative elicitation process, where
use cases are derived from business process models.
As we mentioned before, BORE is designed on
the basis of many existing techniques, ideas,
guidelines and best practices in academia and
industry as well as our experience from previous
projects. Below, we briefly present the work that had
the most impact on our approach.
Dijkman and Joosten (2002) compare the
metamodels of use case diagrams and business
process models. Then, they formally specify a
procedure to transform business process models into
use case diagrams, that complies to this mapping.
Nevertheless, their procedure has some deficiencies.
First, it assumes that all business processes are
supposed to be computer-aided. However, some
processes may simply be more suitable for being
performed by hand. Therefore, in our approach the
cost–benefit analysis precedes the derivation of use-
cases and every elementary business process is
labelled as “A”, “S”, or “M” according to the effect
that the anticipated IS will have on it. Moreover, in
our approach the derivation of use cases is just one
step in the overall process that results in a use-case
model. Second, their procedure does not cater for
ENASE2014-9thInternationalConferenceonEvaluationofNovelSoftwareApproachestoSoftwareEngineering
160

inclusion, extension or generalization of use cases.
As a consequence, the obtained diagrams need
manual reconstruction. Indeed, in the evaluation
case study they conducted in the mortgage
department of a bank, 40% of the generated use
cases had defects. It seems that, even if improved,
the transformation process cannot be done automatic
because it always requires human intelligence.
A similar approach to (Dijkman and Joosten,
2002) can be found in the work of Odeh and Kamm
(2003). They propose a method to explore
relationship between business process model and use
case model. They use Role Activity Diagram (RAD)
to model business. Their method leads to the
derivation of use case model from a process model.
Thus, their and our approach have a similar goal, but
we achieved this goal in a different way, using
different techniques and notations.
The work of Odeh and Kamm (2003) is
continued by Aburub (2012) who propose a four-
steps method: develop a business process model
using RAD model, identify automated activities, link
each business objective with automated activities,
and develop use case model based on objectives and
automated activities.
Štolfa and Vondrák (2004) describe business
process modeling as a tool for definition of
requirements specification. They claim that there are
repeatable situations during transition between
business process modeling and requirement election.
Next, they demonstrate three patterns that can be
applied to support the transition between business
process modeling and other phases of software
process. These patterns can be considered as a
complement to the third stage of our approach.
Vara et al., (2008) propose an approach that tries
to prevent common mistakes detected in practice
such as the lack of understanding of the business by
system analysts, the lack of focus on system
purpose, and miscommunication between business
people and system analysts. Their approach is based
on purpose analysis through BPMN and the MAP
model. Map is a goal/strategy-driven approach to
capture the intentions (goals) of an enterprise or
system and determine the strategies that can
contribute to the fulfilment of these intentions.
Initial BPMN models (As-Is) are updated by the
results of the analysis of the MAP model to get the
To-Be business process models. The models are
validated by end-users, and then analyzed in order to
agree on the effect that the IS may have on the
organizational needs. Finally, requirements are
specified by means of the description of the business
process tasks to be supported by the IS. Our main
objection to their approach is that it introduces
accidental difficulties by using three different
notations (BPMN, Map and UML) during
requirements analysis. Nevertheless, we borrowed
the explicit distinction between As-Is and To-Be
business models from their approach.
BORE also adopts several ideas from EKD
(Bubenko et al., 2001). EKD is an approach that
provides a systematic and controlled way of
analyzing, understanding, developing and
documenting an enterprise and its components. The
EKD Enterprise Model comprises six interrelated
sub-models that describe different aspects of the
enterprise. Links between sub-models make the
model traceable. In contrast to EKD which can be
applied for many different reasons, our approach is
subject to a single well-defined purpose. Therefore
EKD provides means for modeling several aspects
of an enterprise, while our approach concentrates all
efforts on business process models and the way how
these models can be used to support requirements
elicitation.
6 CONCLUSIONS AND
ONGOING WORK
Various methods and techniques have been proposed
for helping make requirements precise, complete,
and consistent. Each of them individually addresses
some problems, but when arbitrary combined with
others may introduce accidental difficulties. In this
study, we demonstrated how to gain synergetic
effect from combining methods and techniques from
different fields of business engineering and
requirements engineering.
The contribution of our research is twofold. On
the one hand, it proposes a structured approach for
deriving system requirements based on business
process models. This approach is especially effective
when system requirements are not fully knowable up
front and must be discovered. Moreover, the built-in
traceability supports impact analysis when an IS
must be adapted to the evolving business processes.
By providing seamless transition from business
process modeling to use case modeling, our
approach can be seen as a further step towards
bridging the gap between business process
engineering and requirements engineering. We also
believe that the application of BORE can address the
number of accidental problems and provide a stable
foundation for attacking the essential difficulties. On
the other hand, our study brings solutions to improve
administrative efficiency of the real organization.
ABusiness-OrientedApproachtoRequirementsElicitation
161

An action research evaluation builds on what
already happens in projects and uses stakeholders’
opinions. In our study, the stakeholders stated that
BORE allowed them to better understand the
business processes and, consequently, the
requirements. Unfortunately, the opinions are flawed
due to the following reasons: (1) the top
management is satisfied because we are
computerizing its organization for free; (2) potential
users are satisfied because the anticipated software
system is likely to lighten their work; (3) we are
satisfied because we achieved our research
objectives. Moreover, the opinions were obtained by
discussing with the stakeholders informally. To
overcome the mentioned weaknesses, we intend to
employ BORE in commercial projects, design a
form to survey these projects, and conduct an
evaluation process using the framework defined in
(Bobkowska, 2005). Future work is also oriented
towards enriching the second stage of our approach
with Business Process Simulation which is
considered as one of the techniques suitable for
discovering process bottlenecks and investigating
business alternatives.
ACKNOWLEDGEMENTS
The study is funded by Polish National Science
Centre, project DEC-2011/01/N/ST6/02752.
REFERENCES
Aburub, F., 2012. Activity-based approach to derive
system models from business process models. In:
International Conference on Information Society,
London, UK
Bendakir, S., El Hamlaoui, M., Zarour, N., 2013. An
agent-oriented approach to support requirements
prioritization processes. In: IEEE International
Workshop on Requirements Engineering 2013,
Constantine.
Bobkowska, A., 2005. A framework for methodologies of
visual modeling language evaluation. In: Proc. of the
2005 symposia on Metainformatics, Esbjerg,
Denmark.
Brooks, F., 1987. No Silver Bullet: Essence and Accidents
of Software Engineering. In: Computer 20, 4 (April
1987), pp. 10–19.
Bubenko, J. A., Persson, A., Stirna, J., 2001. EKD User
Guide, Royal Institute of Technology (KTH) and
Stockholm University, Stockholm, Sweden.
Dijkman, R. M., Joosten, S. M. M., 2002. An Algorithm to
Derive Use Cases from Business Processes. In: 6th
IASTED International Conference on Software
Engineering and Applications, Cambridge, MA.
Eriksson, H., Penker, M., 2000. Business Modeling with
UML: Business Patterns at Work. John Wiley & Sons.
Faulk, S., 1997. Software Requirements: A Tutorial. In:
Thayer, R., Dorfman, M. (Eds.): Software
Requirements Engineering. IEEE Computer Society
press.
Frost, S., Allen, P., 1996. Business Process Modeling,
SELECT Software Tools plc, Cheltenham
Gawin, B., Marcinkowski, B., 2013. Simulating Business
Processes. BPMS and BPMN Standards Applied (in
Polish). Helion.
Gotel, O., Finkelstein, A., 1994. An Analysis of the
Requirements Traceability Problem. In: 1st
International Conference on Requirements
Engineering, Colorado Springs, CO.
Górski, J., 2000. Inżynieria oprogramowania: w projekcie
informatycznym (in Polish). MIKOM.
Hammer, M., 1990. Reengineering work: don’t automate,
obliterate. Harvard Business Review, July-August
1990, pp. 104–112.
Harmon, P., Wolf, C., 2011. Business Process Modeling
Survey. Business Process Trends.
Jacobson, I., 1987. Object-Oriented Development in an
Industrial Environment. In: SIGPLAN Conference on
Object-Oriented Programming Systems, Languages,
and Applications (OOPSLA'87), Orlando, FL.
Jacobson, I., 1992. Object-Oriented Software Engineering:
A Use Case Driven Approach. Addison-Wesley.
Jacobson, I., Booch, G., Rumbaugh, J., 1999. The Unified
Software Development Process. Addison-Wesley.
Leffingwell, D., Widrig, D., 1999. Managing Software
Requirements. Addison Wesley.
Lehtola, L., Kauppinen, M., Vähäniitty, J., Komssi, M.,
2009. Linking business and requirements engineering:
is solution planning a missing activity in software
product companies?. In: Requirements Engineering
vol. 14(2), April 2009, pp. 113–128.
Maciaszek, L., 2005. Requirements analysis and system
design. Addison-Wesley.
Monsalve, C., April, A., Abran, A., 2010. Representing
Unique Stakeholder Perspectives in BPM Notations.
In: 8th ACIS International Conference on Software
Engineering Research, Management and Applications,
Montreal, Canada.
Marcinkowski, B., 2010. Applying Business Process
Modeling Techniques: Case Study. Journal of Internet
Banking and Commerce 15, Array Development.
Odeh, M., Kamm, R., 2003. Bridging the gap between
business models and system models. In: Information
and Software Technology, (45)15, pp. 1053-1060.
Przybyłek, A., 2007. The Integration of Functional
Decomposition with UML Notation in Business
Process Modelling. In: Advances in Information
Systems Development, Vol 1, pp. 85–99.
Przybyłek, M., 2013. Skeletal Algorithms in Process
Mining. In: Studies in Computational Intelligence, vol.
465, pp. 119–134.
Przybyłek, M., 2013. Process mining through tree
automata. In: 5th International Conference on
ENASE2014-9thInternationalConferenceonEvaluationofNovelSoftwareApproachestoSoftwareEngineering
162

Evolutionary Computation Theory and Applications,
Vilamoura, Portugal.
Przybyłek, M., 2013. Dynamic data discovery. In: 8th
International Conference on Knowledge, Information
and Creativity Support Systems, Cracow, Poland.
Ravichandar, R., Arthur, J. D., Perez-Quinones, M., 2007.
Pre-requirement specification traceability: Bridging
the complexity gap through capabilities.
Regnell, B., Andersson, M., Bergstrand, J., 1996. A
Hierarchical Use Case Model with Graphical
Representation. In: IEEE International Symposium
and Workshop on Engineering of Computer-Based
Systems, Friedrichshafen, Germany.
Robertson, S., Robertson, J., 2013. Mastering the
Requirements Process. Addison Wesley.
Sjoberg, D., Dyba, T., Jorgensen, M., 2007. The future of
empirical methods in software engineering research.
In: 2007 Future of Software Engineering (FOSE'07),
Minneapolis, MN.
Štolfa, S., Vondrák, I., 2004. A Description of Business
Process Modeling as a Tool for Definition of
Requirements Specification. In: 12th International
Conf. on Systems Integration, Prague, Czech Republic.
Taylor-Cummings, A., 1998. Bridging the user-IS gap: a
study of major information systems projects. Journal
of Information Technology 13, pp. 29–54.
Torkar, R., Gorschek, T., Feldt, R., Svahnberg, M., Raja,
U., Kamran, K., 2012. Requirements Traceability: a
Systematic Review and Industry Case Study.
International Journal of Software Engineering and
Knowledge Engineering 22(3), pp. 385–434
Vara González, J. L., 2011. Business process-based
requirements specification and object-oriented
conceptual modelling of information systems. PhD
thesis, Universidad Politecnica de Valencia.
Vara, J.L., Sánchez, J., Pastor, Ó., 2008. Business Process
Modelling and Purpose Analysis for Requirements
Analysis of Information Systems. In: 20th
international conference on Advanced Information
Systems Engineering, Montpellier, France.
Weerakkody, V., Currie, W., 2003. Integrating business
process reengineering with information systems
development: issues & implications. In: 2003
international conference on Business process
management (BPM'03), Eindhoven, Netherlands.
Wolf, C., Harmon, P., 2012. The State of Business Process
Management. Business Process Trends.
Wrycza, S., Marcinkowski, B., Wyrzykowski, K., 2005.
UML 2.0 in Information Systems Modeling (in Polish).
Helion.
ABusiness-OrientedApproachtoRequirementsElicitation
163