Information Technology Reference
In-Depth Information
Table
3.
Taxonomy
of
Search
Based
Software
Engineering
Problems
-
OPTIMISATION Perspective
A. OPTIMISATION Perspective
1. Objective Space Dimensionality
(a) Mono-objective
(b) Multi-objective
2. Instance Space Characterisation
(a) Discrete
(b) Continuous
3. Constrained
(a) Yes
(b) No
4. Problem Linearity
(a) Linear
(b) Nonlinear
5. Base NPO Problem Type(s)
(a) Problem Categories as defined in the Compendium of NP Optimisation
Problems
6. Base NPO Problem(s)
(a) Problems as defined in the Compendium of NP Optimisation Problems
statements in the source code, which means that each test case can be represen-
tative of a subset of
S
. Thus, the solutions are set covers for
S
, that is, a subset
of test cases,
C
⊆
C
, such that all statements in
S
are covered, meaning that
each statement is covered by, at least, one of the members of
C
.Thesolution
sought is the one with the lowest cardinality, which will have lowest execution
time, since all test cases have the same execution time.
For the other dimensions in the OPTIMISATION perspective (Table 4), the
Multi-Objective Test Case Selection Problem can be classified as Multi-objective,
having a Discrete instance space, Unconstrained and Linear. Over the SOFT-
WARE ENGINEERING perspective, the problem falls under the Testing/
Validation development stage and is not particular to any specific development
model. Furthermore, it has as main subject descriptor the choice “D.2.5 Testing
and Debugging”, and “Testing Tools” as implicit subject descriptor.
The Next Release Problem (NRP), in its original formulation as a constrained
mono-objective optimisation problem [13], involves determining a set of cus-
tomers which will have their selected requirements delivered in the next software
release. This selection prioritises customers with higher importance to the com-
pany (maximise
i
=1
w
i
), and must respect a pre-determined budget (subject
to
i
=1
c
i
≤
B
).
Search WWH ::
Custom Search