Information Technology Reference
In-Depth Information
2 State-of-the-Art in Automatic SAR
A complete and extensive survey of SAR techniques is proposed by Ducasse
et al. [13] where authors provide an accurate taxonomy of different approaches
according to five distinct aspects, namely the
goals
,the
process
,the
inputs
,the
techniques
and the
outputs
. In this paper, we limit our analysis only to approaches
of automatic SAR for the clustering of functional (sub)modules.
The definition of effective methods to automatically partition systems into
meaningful subsystems, requires that several non trivial issues are considered [26]:
(i)
the level of granularity for the software entities considered in the clustering;
(ii)
the information used to compare software entities, and
(iii)
the clustering
algorithm adopted to group similar artifacts. In Table 1 summarizes the state of
the art regarding software clustering for the recovery of software architectures.
Tabl e 1.
Overview of architecture recovery approaches
Used
Clustering
Auto
m
atic or
Approach
Infor
m
ation
Algorith
m
Se
m
i-auto
m
atic
Anquetil
structural
hierarchical
semi-automatic
and Lethbridge [1]
Mitchell
structural
hill climbing
automatic
and Mancoridis [36]
Doval et al. [12]
structural genetic algorithms
automatic
edge
Bittencourt
betweenness;
semi-
and Guerrero [5]
k-means;
modul. quality;
design structure
matrix
automatic
structural
hierarchical;
semi-
prog. compreh.
Wu
et al.
[47]
structural
automatic
patterns;
Bunch
Tzerpos
structural
hierarchical
semi-automatic
and Holt [43]
Kuhn et al. [29]
lexical
hierarchical
semi-automatic
Risi et al. [38]
lexical
k-means
automatic
Corazza et al.
k-medoids;
lexical
automatic
[8,7,10]
hierarchical
Maqbool
lexical
hierarchical
semi-automatic
and Babri [34]
structural
Maletic
lexical
minimum
semi-
and Marcus [33]
structural
spanning tree
automatic
lexical
Scanniello et al. [42]
k-means
automatic
structural
Search WWH ::
Custom Search