D.3.3 ALGORITHMS FOR INCREMENTAL ... - SecureChange

Dealing with Known Unknowns: A Goal-based Approach 9
(a) AMAN Evolution 1
(b) AMAN Evolution 2
The right hand corner of the two models corresponds to the different evolutions of the
model that we have described in Example 2.
Fig. 2 Alternative Evolutions for AMAN Deployment.
a subpart of a goal model is an arbitrary goal and all its directly and indirectly children.
The evolutionary goal model, therefore, is a tuple of an original goal model
and a set of observable rules, 〈EM, R o〉.
It is not feasible to represent all evolutions in the same diagram. In order to understand
the level of complexity we show in Fig. 2 two alternative models in which
a number of observable evolutions rules have been triggered. In comparison with
Fig. 1 the new goals have been marked in green. It is clearly impossible to identify
the two simple evolution rules that we have described. These evolutions must be
represented as local evolutions of the corresponding fragment of the models.
Here we outline how evolution rules are visualized in an evolutionary goal
model, more discussion can be found in a technical report [37]. Fig. 3 illustrates
two graphical representations of an observable rule, tree-like and swimlane-based,
taken from Example 2.
Fig. 3(b), meanwhile, describes the swimlane-based representation where original
model is on the left side, and all the possible evolutions are located next to the
right. Evolution probability is indicated by the label and the shade of swimlanes.