Authors:
João Choma Neto
;
Thelma E. Colanzi
and
Aline M. M. Miotto Amaral
Affiliation:
State University of Maringá, Brazil
Keyword(s):
Software Product Line, Product Line Architecture Design, Memetic Algorithm.
Related
Ontology
Subjects/Areas/Topics:
Artificial Intelligence and Decision Support Systems
;
Enterprise Information Systems
;
Evolutionary Programming
;
Information Systems Analysis and Specification
;
Software Engineering
Abstract:
Basic design principles, feature modularization, and SPL extensibility of Product Line Architecture (PLA)
design have been optimized by multi-objective genetic algorithms. Until now, memetic algorithms have not
been used for PLA design optimization. Considering that memetic algorithms (MA) have achieved better
quality solutions than the solutions obtained by genetic algorithms (GA) and that previous study involving the
application of design patterns to PLA design optimization returned promising results, bringing the motivation
in investigating the use of MA and the Design Pattern Search Operator as local search to the referred context.
This work presents an exploratory study aimed to characterize the application of using MA in PLA design
optimization. When compared with a GA approach, the results show thatMAare promising, since the obtained
solutions are slightly better than solutions found by the GA. A pattern application rate was identified in about
30 % of the solutions o
btained by MA. However, the qualitative analysis showed that the existing global search
operators need to be refactored for the joint use with the MA approach.
(More)