A Comparative Study of Software Requirement, Elicitation, Prioritization and Decision Making.

Pradeep Kumar. G.et al. Int. Journal of Engineering Research and Applications www.ijera.com
ISSN: 2248-9622, Vol. 6, Issue 4, (Part - 5) April 2016, pp.01-07
www.ijera.com 1|P a g e
A Comparative Study of Software Requirement, Elicitation,
Prioritization and Decision Making.
Pradeep Kumar. G 1
, Sandhiya .R 2
, Prof. Manjula. R3
.
1, 2, 3
(Department of Computer Science, VIT University, Vellore- 632014)
ABSTRACT
The failure of many software systems are mainly due to the lack of the requirement engineering. Where
software requirement play a very vital role in the field of software engineering. The main task of the
requirement engineering are eliciting the requirements from the customer and to prioritize those requirements to
make decisions in the software design. Prioritization of the software requirement is very much useful in giving
priority within the set of requirements. Requirement prioritization is very much important when there are strict
constraints on schedule and the resources, then the software engineer must take some decisions on neglecting or
to give prioritization to some of the requirements that are to be added to the project which makes it successful.
This paper is the frame work of comparison of various techniques and to propose a most competent method
among them.
Keywords: software requirement prioritization, dynamic, prioritization, stakeholders, AHP, Hundred-dollar
test, Ranking, Numerical assignment.
I. INTRODUCTION
Requirements engineering is found to
handle the activities that challenge to know the
precise requirements of the clients in a software
management system and to convert such needs into
exact and definite statements, which are then
utilized in the system growth.
In this paper we are using AHP method, which was
established for performing elicitations,
prioritization and decision making processes. It is
importantly known for optimizing the decision
making when one is facing with a mixture of
quantitative, qualitative and occasionally
conflicting factors that are in the concern. In
general Elicitation involves defining the
requirements of the stakeholders and
understanding, finding, mining and /or determining
requirements of the clients and the stakeholders [1].
Requirement elicitation process is one among the
most significant knowledge intensive activities of
the software development. A study that has been
performed by [2] states that, 70% of the system
errors are because of the insufficient specification
of the system and 30% of the system errors are
because of the design issues. Analyzing the
software security system based on the requirements
of the system elicited that are in the form of use and
misuse cases. Use cases are established to be
helpful in the elicitation of communication about,
and the function requirements documentation.
Using the elicitation process can assist in
generating a steady and whole rest of the security
requirements. The last but most important step is
prioritizing the software, which mainly relies on the
Requirements and the elicitation factors. In
this paper we make a comparative study among few
prioritization methods and conclude that which
method can yield feasible solution.
II. RELATED WORKS
Software requirement prioritization is
made by comparing AHP and ELECTRE I. In
order to reflect the reality several stake holders and
software professionals are directly involved in the
case study. AHP method is difficult in performing
software projects with a huge volume of
requirements. Thus it is concluded that ELECTRE
I, is more appropriate than AHP [1]. An
Experimental pattern is made to compare three
Software Requirements Prioritization Techniques
that are AHP, Cumulative Voting, and Numerical
Assignment. The main goal of this pattern is to
analyse which technique is best suitable in
performing, software prioritization in terms of time,
accuracy, and portability. And finally it is
concluded that cumulative voting is found to
produce good prioritization results. A machine
learning approach is carried out for performing
software requirement prioritization. A method
called Case-Based Ranking (CBRank) is used, that
associates the preferences of project stakeholders
along with the approximations of ordering that is
computed with the help of machine learning
approaches and brings out with promising
advantages. In the first place, the human push to
input preference data is decreased, while saving the
exactness of the last positioning appraisals. Second,
space information encoded as incomplete request
relations characterized over the prerequisite
RESEARCH ARTICLE OPEN ACCESS

qualities can be misused, in this manner supporting
a versatile elicitation process. Exact assessments of
properties of CBRank are performed on reproduced
information and contrasted and a best in class
prioritization strategy, giving confirmation of the
technique capacity to bolster the administration of
the trade-off between elicitation exertion and
positioning precision and to endeavor area learning.
At long last, the CBRank technique has been
situated regarding best in class approaches, with
specific reference to the AHP strategy, which can
likewise be viewed as an occasion of the case-
based critical thinking worldview .Performing
software requirement elicitation and prioritization
using a process called analytical hierarchy. Success
and the failure of a software system mainly rely on
the quality of the requirements. The techniques that
are hired during the requirements elicitation
influence the requirements quality. Elicitation of
Requirements is the most crucial part of the
software development, because the defects at this
initial stage is get transmitted through the
development process and it is hard to repair them
later. An algorithmic method is carried out to elicit
the software requirements prioritization using
Analytic Hierarchy Process (AHP), by which the
requirements are ranked and implemented in a very
easy way. Software requirement prioritization is
carried out using fuzzy multi attribute decision
making, which overcomes the defects like lack of
consistency, time complexity, and implementation
difficulties. This approach considers the indefinite
nature of the quality attributes and requirements by
designing the end as a fuzzy variable. This
requirement prioritization problem is designed as a
fuzzy multi attribute decision problem, by which
predictable operator value is used in order to rank
the alternatives that are listed in the formulation of
a problem. Finally it has been stated that this
approach is capable of breaking several features
and criteria for providing a complete view of
quality requirements. A new way for handling
requirement prioritization using B-tree method is
carried out. An organization must know about its
customer perspective towards their products, as the
customer expectations are not static, periodic
prioritization is a must in order to increase the
customer satisfaction. When considering a larger
requirements case the prioritization technique
requires more comparisons, thus B-tree method is
used in order to overcome such problems. From
this method it is found that the requirement
comparison is drastically reduced if the branching
factor is high, where its control is given to the
prioritization group. These can be managed by
simply altering the t value. Thus B-tree method is
used in which the comparison required by the
respective method is drastically reduced.
Classifying several software requirement
prioritization methods by highlighting their
important features is carried out. This requirement
prioritization process should be simple and firm
producing us with valid results. The framework
involved in developing the design to describe the
proposals of the requirements and these
prioritization methods are analyzed using a
conceptual framework. The differences among
these methods are highlighted by stressing their
vital features which are useful in the elicitation
process. Optimization of Requirement
Prioritization using a new technique called
FuzzyHCV is carried out. This FuzzyHCV is a
hybridized method of Fuzzy Expert System and
Hierarchical Cumulative Voting (HCV).
Comparative analysis is made between the
proposed technique and the currently available
HCV techniques and found that the proposed
FuzzyHCV technique is more consistent than any
other existing techniques. Performing and
analyzing an effective test case prioritization
agenda, which make use of design diagrams,
software requirements specification, test cases and
source codes as input and delivers a prioritized
order of test cases as an output. The requirement
IDs are fragmented into words in order to calculate
the requirements relativity and the calculation to
find the interconnection of the activities, the design
drawings were extracted in the XML format. Final
weight is calculated by multiplying design
interconnectivity, Requirements relativity and class
dependencies with their assigned weights. The test
cases are selected by mapping the test cases and
customers’ requirements using that weight. This
proposed agenda is analyzed and finally stated that
the collaborative information during the
prioritization process is beneficial. A hybridized
approach is provided for requirement engineering
in the agile software development, with the
assistance of JAD and prioritization method.
Prioritization is performed with the help of
viewpoint, by which increment selection is made
very easy. Thus it is clear that the hybrid approach
strongly focus on increased delivery and quality of
the software.
III. ANALYTICAL HIERARCHY
PROCESS (AHP)
The analytical hierarchy process which
was developed by Thomas L. Saaty in the period of
1970s. It is a structured Techniques for organizing
and analyzing complex decisions, from then many
more studies and refinement were carried out. It is
most widely used to make group decisions and in
some of the decision situations. It is also very much
used in other fields such as education, business,
healthcare, industry and shipbuilding. AHP helps

the decision makers to find the decision that is best
suitable, rather than choosing the correct decision
for their goal. It also gives the perfect structure to
the decision problem by providing a comprehensive
and rational framework that helps in better
representation and quantification of its elements
and making relationship with the elements to
overall goals. It is also useful in evaluation of
alternative solutions.
The first step in the AHP is to decompose
the decision problem into hierarchy of divided sub-
problem and each of the sub-problem can be
analyzed independently. After the hierarchy is
built, the decision makers evaluate the various
elements by comparing them to each other two at a
time, in systematic approach. AHP converts theses
evaluations into numeric values and that are
processed and compared to the entire range of the
problem. In the final step, for the each of the
decision alternatives numerical priorities have been
created.
IV. REQUIREMENTS ELICITATION
TECHNIQUES:
The most important task in the software
development is to extract the requirement from the
user, here many techniques are developed for
eliciting requirements from the customer.
Development approach in the some of the
elicitation techniques mainly concentrate on the
security issues. On the other hand, which is
traditional development to extract the requirements
from the customer in general, this also support the
security issues. In general software requirement
elicitation is performed with the use of any
elicitation methods or with help of few techniques.
There are many elicitation techniques that helps in
the extraction of requirement from the customer,
few of them will be discussed for this study.
Elicitation of the requirement is a method
in which the requirements of developing a software
are collected very carefully which influences the
development of the software and its reliability and
feasibility. For this reason, requirement elicitation
is given highest priority during the initial stages of
the development of software. When this part is
ignored or proper work is not put into it, it affects
the overall work greatly. Any mistakes that are
done at this stage or any constraints that are missed
out will be infectious even in the later stages of the
development, which may possibly negate a lot of
other good work put into it. When considering the
failures of software, 70% of them is because of the
improper requirement elicitation.
There are lot many approaches which can
be used for the requirement elicitation. Among
them two techniques are very interesting one. They
are
1. Misuse Cases
2. Joint Application Development (JAD)
Misuse Cases
This technique uses the uses cases to
identify the defects or hidden constraints of the
software development. It is a kind of backward
engineering, where an undesirable scenario or
constraint of the development is identified first,
rather than setting up the constraints for the end
user requirement. This technique excels in several
environment, where there are several unknown
aspects that can affect the system, which make
them very useful in the security requirement
elicitation. This technique creates a tree structure
with the parent nodes reflecting the requirement
and the child nodes corresponds to identifying the
constraints. Whenever the child identifies an
anomaly or a flaw, it propagates the details to the
parent node for it to make necessary changes and
consider them as a constraint. This makes the
technique as a bottom up approach, eliciting the
requirements from the details identified in the
system.
JAD
Joint Application development is a very
effective method for satisfying the end user and the
shareholders. Here all the major parties involved in
the project, including core developers, managers,
end users and other interested parties. Here all the
requirement elicitation and constraints are shared
with the users and their opinions on the matter are
gathered. Here no technical aspect of the
development are used as a talking point. This will
provide an end result that satisfies the consumer
greatly. But it is highly mandatory to initially
perform a few tasks ahead of the meeting like
gathering the technical aspects, collecting initial
requirements from it and features that can satisfy it.
These two techniques are comparatively
very different that can be used together for different
purposes. The Misuse Cases are arguably the best
in terms of identifying the security threats while the
JAD focuses on the satisfaction of the end user.
Understanding the software and its usage is
important in choosing the elicitation method.
Now that all the requirements are
collected, whether by JAD or Misuse Cases, an
impending task of prioritizing them arises. This is
handled by the AHP. This is a decision making
mechanism that identifies the right kind of
requirement using pair wise comparison matrix that
has to be focused. This technique greatly makes the
software more desirable and reliable. The
developers can make verifications or changes in the
result obtained by the AHP, which makes them
more effective.

V. PRIORITIZATION TECHNIQUES
Analytical Hierarchical Process (AHP):
The AHP method was discovered by Saaty
[4] in the year 1980. The study stated by Regnell et
al [12] declare that AHP a very hopeful method,
but its unable to adapt to the prioritization
environment that is found with several
stakeholders, thus it should be designed in some
other way to adapt the situation that is given. Still
no research has been carried out to state what kind
of modification could function properly.
In case of AHP method, requirements of
the candidates are analyzed in a pair wise manner,
and also to what degree one requirement is highly
significant than other one. As this method describes
the pair a wise analysis that is carried out on all the
requirements of the candidate, the count of
comparison raises polynomial. Pair wise analysis
is made for a system that has n requirements which
is represented as, n. (n - 1)/2.
Cumulative Voting (CV):
Cumulative Voting (CV) was discovered
by Widrig and Leffingwell , it is a linear and
interesting voting system in which each and every
stakeholder is provided with a constant quantity of
invented units that a person could make use of it for
voting purpose in goodwill of some significant
issues [14]. By this path the quantity of money that
has been allotted to the issue corresponding to the
relative preference of respondent with other issues
related to it. The statements are scattered according
to the stakeholder’s preference. Possibly the
stakeholders can divide the amount equally to many
issues.
CV method is also called as “proportional
voting” method, it is because the quantity of units
that is assigned to the issues provides the
information about the virtual priority of some exact
problems that is related to some other problems.
The word “proportional” replicates the main truth
that if the quantity of units that has been allotted to
a specific issue is found separated by some constant
units that is presented to each and every
stakeholders, the final outcome turn out to be a zero
and one proportion. Thus the ratings of the
stakeholders for a certain layout of issues are thus
taken as the mixture of person’s perspective in case
of any problems, where each and every issue takes
a certain amount of choices into the person’s
preferences.
This process could result in some
problems that are given with zero units, which
states that the particular stakeholder takes these
problems as an insignificant one. In particular type
of data the value zero is generally a problem it is
because they prepare the virtual needs or significant
as entirely a meaningless one making the
calculation of zeros as impossible one. In common
the rule of CV method allows the stakeholders to
distribute their entire quantity without any
limitations.
Planning Game (PG):
In the case of extreme programming
requirements are given on a story card by the
customer, which differ based on several factors.
Thereby the requirements are separated into three
piles by the respective customer. Beck states that
piles should have certain names that are nice to
hold on [16]. Thereby the programmer calculates
the time that is taken by each requirement for
implementing and arranging the requirements into
three various piles.
The final resulting of this sorting process
is placing the sorted requirements on the ordinary
scale. As the PG process consumes single
requirement and concludes the pile to which the
requirement has to be placed, thus time taken for
prioritizing n number of requirements is considered
as n comparisons. It clearly states that PG method
is very supple and can be performed on several
requirements with less time consumption.
Binary Search Tree (BST):
BST is an algorithm which is mainly to
store the data and to retrieve back for the future
reference. In the binary search tree, the tree which
will be empty or it is engaged with the two child
nodes. There will be three node, they are root node,
right node and left node. In which right node that
contains value which is highly important than the
value of root node, where the value of left node
which is less important when compared to root
node. Each child node which acts as the root node
for its child in the next level of hierarchy, where it
is known as leaf. This method helps in the recursive
search in this algorithm. The merits of using binary
search tree algorithm is that, when the requirements
are inserted in the order it takes only n log n
number of comparisons for the prioritization of the
software requirements. With help of this BST can
prioritize in faster manner, where it can prioritize
more number of requirements in the given time. It
can be scaled up for more number of requirements.

Experimental work
Experimental work has been carried out
by the comparison of various factors with the
prioritization techniques such as ease of use,
scalability, total time taken, total number of
comparisons etc.
Ease of use
Fig: represents the parameter of ease of
use, here comparisons between various
prioritization have been carried out and the result is
PG is more ease of use than any other techniques
compared here.
Total time taken
Fig: represents the parameter total time
taken, here AHP is more time taken than any other
methods compared here.
Accuracy
Fig: represents accuracy measures of the
prioritization process. This graph represents that
BST is more accurate than all other comparisons.
Scalability
Fig: represents scalability measure of the
techniques, BST is more scalable when compared
with the above techniques.
Total number of comparisons
Fig: representation of total number of comparison
needed to show the results.
Overall score
Fig: shows that PG is more suitable than any other
techniques among the above comparison.
VI. DECISION MAKING
This section is all about analytic hierarchy
process (AHP) which is one of the well-known
methods of decision making for a multi- criteria
decision problems. AHP is more suitable when
there is a mixture of qualitative and quantitative
aspects and when there are more clash among
several requirements which are to be taken as
deliberations. AHP is developed in such a way that,
to enhance the role of decision making. AHP is
more effective in the case of irreparable decisions.
The decision-making process comprises the use of
perception, intelligence and creativeness of the
humans by satisfying their fundamental needs.

Estimating a decision involves various concerns
like welfares are extracted for making the proper
decision, risks, costs, and profit arising from the
actions that are taken if the decision is made wrong.
Methods of Decision making ranges from
dependence on chance to the structured decision
making tools. Considering all the features of the
element, which makes decision making more
effective. The merits and demerits of the object are
to be taken into consideration that’s helps in the
upcoming achievements and to be in the business
life or the life can be enhanced. In recent days of
decision making has been more complex when
many feature are to be taken into consideration.
AHP was developed to overcome these issues like
ranking, evaluation of decision and to prioritize the
requirements.
In the year 1970s, AHP was developed by
Thomas Saaty, who was working as a professor in
the Wharton School of Business institute and as a
consultant in an agency called Arms Control
Disarmament. There he faced with some
problematic factors like higher cost and a host of
the considerations that are not specified easily. The
AHP method deals with factors like resource,
weapons tradeoffs, asset allocation, and decision
making, where the judgements of the decision
makers are made used in order to break down the
difficulties into hierarchies. The complexity of the
problem is denoted by the count of levels that are
presented in the hierarchy that join with the model
of the judgement maker’s problem that has to be
solved. Hierarchy is made used in order to extract
the ratio values for project risks, the related
measure which are against the goals of the
organization (organizational effects, satisfaction of
the customer, service/product, human resources and
finance) and decision replacements. This AHP
method make use of matrix algebra for arranging
the factors in order to attain optimal solution and
thus it is used in forming several trillion dollar
decisions.
AHP decomposes the issues into
hierarchies with the help of sentences of the
decision makers. The number of stages in the
hierarchy is the representation of the complication
in the problem, the problem to be cleared here AHP
combines with the decision maker’s model. The
levels of hierarchy which is used to eradicate the
measures of ratio-scaled to which to make some
alternative in the decisions and also the measure of
the relative values by making the alternative
decisions which affects the project risk and for the
organizational goals. Mathematically the optimal
solution is derived with help of matrix algebra in
AHP. AHP which is said to be time-tested method
that can also be used in multibillion dollar
decisions.
AHP consist of 4 steps
1. Outlining of the problem and stating of goal or
objective.
2. Expression of factors those are influencing the
goals. And structure these factors into
hierarchical levels.
3. Using of paired association of each of the
factor with respect to each other that forms the
comparison matrix.
4. Synthesize the alternative decisions rank until
the final choice is made.
VII. CONCLUSION
Comparison has been made between the
techniques that are used at the different stages of
the software engineering. They are elicitation,
prioritization and decision making of the software
requirements. The conclusion from this paper work
is that AHP is well suitable technique for the
software requirement elicitation and for the
decision making for the software requirements. But
then for the process of requirement prioritization
AHP is not suitable because of it takes more
number of comparison i.e. n.(n-1)/2 number of
comparisons. This time complexity need to be
reduced and that makes AHP as an effective
technique for the software requirement
prioritization.

REFERENCE
[1]. Fernandez, J. M., Prozil Rodrigues, S., &
Costa, L. A. (2015, June). Comparing
AHP and ELECTRE I for prioritizing
software requirements. In Software
Engineering, Artificial Intelligence,
Networking and Parallel/Distributed
Computing (SNPD), 2015 16th
IEEE/ACIS International Conference
on (pp. 1-8). IEEE.
[2]. Mustafa, B. A., & Zainuddin, A. (2014,
August). An experimental design to
compare software requirements
prioritization techniques.
In Computational Science and Technology
(ICCST), 2014 International Conference
on (pp. 1-5). IEEE.
[3]. Perini, A., Susi, A., &Avesani, P. (2013).
A machine learning approach to software
requirements prioritization. Software
Engineering, IEEE Transactions on, 39(4),
445-461.
[4]. Sadiq, M., Ghafir, S., &Shahid, M. (2009,
October). An approach for eliciting
software requirements and its
prioritization using analytic hierarchy
process. In Advances in
RecentTechnologies in Communication
and Computing, 2009. ARTCom'09.
International Conference on (pp. 790-
795). IEEE.
[5]. Ejnioui, A., Otero, C. E., & Qureshi, A. A.
(2012, October). Software requirement
prioritization using fuzzy multi-attribute
decision making. InOpen Systems (ICOS),
2012 IEEE Conference on (pp. 1-6). IEEE.
[6]. Beg, R., Abbas, Q., &Verma, R. P. (2008,
July). An approach for requirement
prioritization using b-tree. In Emerging
Trends in Engineering and Technology,
2008. ICETET'08. First International
Conference on (pp. 1216-1221). IEEE.
[7]. MartínezCarod, N., &Cechich, A. (2005).
Classifying software requirement
prioritization approaches. In XI
CongresoArgentino de Ciencias de la
Computación.
[8]. Sharif, N., Zafar, K., &Zyad, W. (2014,
April). Optimization of requirement
prioritization using Computational
Intelligence technique. In Robotics and
Emerging Allied Technologies in
Engineering (iCREATE), 2014
International Conference on (pp. 228-
234). IEEE.
[9]. Siddik, M. S., &Sakib, K. (2014,
December). RDCC: An effective test case
prioritization framework using software
requirements, design and source code
collaboration. In Computer and
Information Technology (ICCIT), 2014
17th International Conference on (pp. 75-
80). IEEE.
[10]. Kumar, M., Shukla, M., & Agarwal, S.
(2013, December). A Hybrid Approach of
Requirement Engineering in Agile
Software Development. In Machine
Intelligence and Research Advancement
(ICMIRA), 2013 International Conference
on (pp. 515-519). IEEE.
[11]. Ann M. Hickey, Alan M. Davis,
“Requirements Elicitation and Elicitation
technique selection: A Model for Two
knowledge-Intensive Software
development Process”, Proceedings of the
36th IEEE International Conference on
System Sciences, 2002.
[12]. Beichter F et al, “ SLAN-4-A Software
Specification and Design Language”,
IEEE transaction on Software
Engineering, SE- 10,2, 1994, pp 155- 162

A Comparative Study of Software Requirement, Elicitation, Prioritization and Decision Making.

More Related Content

What's hot (20)

Similar to A Comparative Study of Software Requirement, Elicitation, Prioritization and Decision Making. (20)

Recently uploaded (20)

A Comparative Study of Software Requirement, Elicitation, Prioritization and Decision Making.