Journal of Software Engineering and Applications, 2011, 4, 345-349
doi:10.4236/jsea.2011.46039 Published Online June 2011 (http://www.SciRP.org/journal/jsea)
345
Improvement of Software Quality Attributes in
Object Oriented Analysis and Design Phase Using
Goal-Question-Metric Paradigm
Hitesh Rajput1, Lalit Kumar Singh2
1Product Engineering Services, Patni Computers, Navi Mumbai, India; 2Department of Atomic Energy, Nuclear Power Corporation
of India Ltd, Bhabha Atomic Research Centre, Mumbai, India.
Received April 19th, 2011, revised May 16th, 2011; accepted May 24th, 2011.
ABSTRACT
In a competitive busine ss landscape, large organizations such as insurance companies and banks are under high pres-
sure to innovate, improvise and distinguish their products and services while continuing to reduce the time-to market
for new product introductions. Traditional approaches to software reliability modeling for such software are black
box-based. Bad structure or model again can lead us to lower down these non functional properties. The basic con-
structs of the model are objects. We will not deal about the identification of the objects, as may be referred in many
books, but how to model those objects. The objective of this paper is to provide an philosophical approach, using
Goal-Question-Metric paradigm, to structure or model the identified objects of software system, in better way to im-
prove the quality of th e software.
Keywords: Software Reliability, OOAD, Class, Object, Association, GQM, MFC
1. Introduction
Many companies invest significant resources and effort
in software development. Because of today’s high market
demands for software; its size, complexity, qu ality needs
are increasing rapidly. Due to which the various prob-
lems, specific for software development, viz planning
difficulties, unknown or bad quality, projects are never
ended, milestones that are reached months or years too
late, or developers who are working mostly unstructured,
under high stress; are also increasing. Those problems
are generally known as “software crisis” and are being
tackled by ‘Software Process Improvement’ (SPI) [1].
The object-oriented paradigm gives us tools for fighting
the software crisis: Object-oriented techniques make it
possible for us to handle large systems, change them,
reuse parts of old systems in new systems, ease the
communication between customer and developer, and
much more. When we are fighting the software crisis we
are fighting human nature and human inability to handle
complexity. The performance/price ratio has exploded
for hardware, whereas the increase in the performance/
price ratio is hardly noticeable for software.
Worse than this quantity problem is the quality pro-
blem. The quality problem is due to lack of SPI. One of
the ways to handle is problem is through object-oriented
modeling. The act of making model is known as model-
ing. A model is a representation of a real-world process,
device or concept [IEEE-729]. When we start building a
system, we might find tangible and non tangible items
(ideas, etc). All these must be put in our model. So, mod-
el is the collection of real objects that exist in the world
by applying our ideas. This can be understood with the
help of Figure 1. Basic concepts used in object-ori-
ented modeling are objects, classes, attributes and mes-
sages and important concepts such as information hiding.
We thus need a technique that can be used to arrange our
model so we can look at it both in detail and, when
needed, in larger chunks, hiding its internal details. The
mind mapping technique is an excellent example of how
good structures can in crease our ability to manag e a larg e
number of details.
We find lot of things when modeling but what do we
find? Purpose is a key. Utility is in the eye of beholder.
We cannot tell if we are going in the right direction until
you know where we want to go. So we propose to make
the use of Goal-Question-Metric approach to model the

Improvement of Software Quality Attributes in Object Oriented Analysis and Design Phase
346 Using Goal-Question-Metric Paradigm
Figure 1. Real objects and ideas are put in the model.
objects or our system(in broad sense).
The GQM approach to process and metrics, first sug-
gested by Basili and his colleagues, has proven to be a
particularly effective approach to selecting and imple-
menting metrics [Basili and Weiss, 1984; Basili and
Rombach, 1 98 8 ].
The GQM paradigm is explained in Section 2.
This paper is organized as follows: in Section 2, GQM
paradigm is explained. Section 3 focuses on the way to
model the reality, based on GQM. Section 4 concludes
the paper.
2. The GQM Paradigm
Many metrics programs begin by measuring what is
convenient or easy to measure, rather than by measuring
what is needed. Such programs often fail because the
resulting data are not useful to the developers and main-
tainers of the software. A measurement program can be
more successful if it is designed with the goals of the
project in mind. The GQM approach provides a frame-
work involving three steps [2,3]:
1) List the major goals of the development or main-
tenance projec t.
2) Derive from each goal the questions that must be
answered to determine if the go als are being met.
3) Decide what must be measured in order to be able
to answer the questions adequately.
The GQM paradigm provides a method for top-down
metric definition and bottom-up data interpretation (Fig-
ure 2).
3. Relationship of GQM and Model of the
Identified Objects i n Th eo re ti cal
Perspective
We know the various methods to identify the potential
objects like Use case text, CRC models, etc. Once we
find the potential objects, to model them, we demonstrate
that Goal-Question-Metric will be able to help in this
regard. We must follow the following steps:
1) What is our purpose-our intention? Our purpose is
obviously to find the potential objects. So the GOAL is
to model the potential objects.
2) Now we should ask various set of QUESTIONS
like what are the requirements of our system, what are
Figure 2. The Goal-Question-Metric paradigm.

Improvement of Software Quality Attributes in Object Oriented Analysis and Design Phase 347
Using Goal-Question-Metric Paradigm
the possible ways through which objects can be shown or
in what possible forms the objects can be shown. These
questions can be answered with the help of requirements
specification or analysis models.
3) Now we can select the required form of the objects
with the help of domain knowledge or requirements spe-
cification.
4. Applying GQM to Model the Reality in
Object Form
We feel that the GQM paradigm can be adequately used
to model (object derivation) the reality because the real-
ity can be modeled in number of ways, depending on the
Goal. This can be illustrated with the help of following
two examples:
Example 1 [3]. Nature of fundamental particles likes
electrons, protons, neutrons, etc. It became palpable that
the particles did not always behave as something that had
a well-defined shape and size. Apparently, the funda-
mental particles in some experiments behaved as parti-
cles with characteristics like mass and electric charges,
where as in other experiments they behaved as waves
producing diffraction patterns (see Figure 3). The model
needed to explain an experiment depended on the intent
or goal of the experiment.
Example 2. Classification of Nuclear Power Plant
(NPP)-the people who live near to NPP will classify it in
one way (probably launch complaint about the radiation
exposure or machinery noise). The people who use its
electricity will classify it in some different way (probably
like a dynamo without their own control). The people who
are into maintenance of NPP will classify it in other way.
When we try to describe computer systems, so various
purposes also yield various models. We always have to
define a clear goal, otherwise modeling situation will
soon become very awkward indeed. Let us assume that
in Nuclear Power Plant there is a Nitrogen Tank to cre-
ate some Pressure, may be for process requirement. Be-
ing an important object, NPP Engineering group also
store many tanks in the store. Now this situation can be
modeled by GQM paradigm as in Figure 4. If the pur-
poses of modeling nitrogen tank because we need it to
pressurize the system, then it is better to model two dif-
ferent simple objects rather than one complex object,
else to model only one complex object is sufficient (see
Figure 5). we shall store a tank in a store is very differ-
ent from the purpose of modeling a nitrogen tank be-
cause we need it to pressurize the system, then it is better
to model two different simple objects rather than one
Figure 3. An electron can behave like a wave or a particle
depending on goal.
Figure 4. Use of GQM to model the object (Nitrogen Tank) of NPP.

Improvement of Software Quality Attributes in Object Oriented Analysis and Design Phase
348 Using Goal-Question-Metric Paradigm
Figure 5. (a) One way to model nitrogen tank, (b) Other
way to model nitrogen tank.
complex object, else to model only on e complex obj ect is
sufficient. The above concept has also been implemented
in Document/View Architecture [4] in (Microsoft
Foundation Classes) MFC. The MFC multiple-document
interface (MDI) architecture is focused on opening multi-
ple documents of the same class. The document class
contains the internal representation of the application
data. For each new or opened document the framework
creates an instance of Cdocument. The frame class de-
scribes the user interface of document windows of the
application, typically MDI frames. The view class sh ow s
a [5] graphical representation of the document type. All
classes are as CObject subclasses. In MFC the CDocu-
ment based classes act as observers for the views at-
tached to them. Whenever user modifies document data
via one of the views the view notifies the associated
document by calling its UpdateAllViews() member. The
document then notifies all views that are attach ed to it by
calling their OnUpdate() member. The views can then
redraw their data based on the hints passed to them as
parameters (see Figure 6).
The above method is ju st like a brainstorming in which
our set of questions is a tool to make a discussion take a
new and different direction.
5. A Comparative Study in between the
Classical Approach and the Proposed
Philosophical Approach
The classical approach to model the objects is very nicely
given in the text book by Pressman [6], in which a data
object can be treated an external entity, anything that
produces or consumes information, a thing like report, an
occurrence like telephone call, event, a role, a place or a
structure. It has been stated that according to the re-
quirements the data object should be modeled but no
technique or model is given for that. In this case a mod-
eler can model the object which may lead to many
changes in the future, may be during development phase.
Figure 6. Observer in document/view model.

Improvement of Software Quality Attributes in Object Oriented Analysis and Design Phase 349
Using Goal-Question-Metric Paradigm
So modeler may have to re-model the objects which
could be very costly and time consuming, which can slip
off the deadline. As we have seen above, our method
provide a clear cut method to model the object during the
analysis or design phase to overcome the flaws, using the
traditional method.
6. Conclusions
The concept of Goal-Question-Metric paradigm [7-8] has
been proposed to model the identified objects, which are
the basic constructs of model, of the software system, in
better way for the improvement of non functional proper-
ties, which are the quality attributes of software. Two
examples have been used for illustratio n.
REFERENCES
[1] S. Sigfried, “Understanding Object-Oriented Software
Engineering,” IEEE Press Understanding Science &
Technology Series, Piscataway, New Je rsey, 1995.
[2] N. E. Fenton and S. L. Pfleeger, “Software Metrics, A
Rigorous and Practical Approach,” 2nd Edition, Interna-
tional Thomson Computer Press, Boston, 1996.
[3] R. V. Solingen, et al., “Improvement by Goal-Oriented
Measurement,” Proceedings of European Software Engi-
neering Process Group Conference, 16-20 June 1997,
Amsterdam, The Netherlands, p. 11.
[4] SDI (Single Document Interface) and MDI (Multiple
Document Interface) Documents, “Views and Framework
(MFC),” 2005.
http://www.msdn.microsoft.com/en-us/library/b2kye6c4(
v=vs.80).aspx
[5] Multiple Document Interface, “Multiple Document In-
ference,” August 2007.
http://www.en.wikipedia.org/wiki/ Mul-tiple_d ocument_in-
terface
[6] R. S. Pressman, “Software Engineering: A Practitioner’s
Approach,” McGraw-Hill Science Engineering Publica-
tion, New York, 2004.
[7] L. Cyra, et al., “Extending GQM by Argument Struc-
tures,” International Federation for Information Proc-
essing (IFIP), Vol. 5082, 2008, pp. 26-39.
[8] F. V. Latum, R. V. Solingen, M. Oivo, B. Hoisl, D.
Rombach and G. Ruhe, “Adopting GQM-Based Meas-
urement in an Industrial Environment,” Software IEEE,
Schlumberger RPS, Vol. 15, No. 1, 1998, pp. 78-86.