A Rationale-based Model for Architecture Design Reasoning
A Rationale-based Model for Architecture Design Reasoning
A Rationale-based Model for Architecture Design Reasoning
Create successful ePaper yourself
Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.
6.2. <strong>Architecture</strong> <strong>Rationale</strong> and Elements Linkage (AREL)<br />
∀r ∈ AR, there exists a cause e ∈ AE such that (e, r) ∈ P L and an effect e ′ ∈ AE<br />
such that (r, e ′ ) ∈ P L;<br />
2. no subset of links in P L <strong>for</strong>m a directed cycle.<br />
According to Definition 1, an AR is connected to a minimum of two AE nodes through<br />
links, one from a cause AE and one to an effect AE. As such, the links to the AEs then<br />
represent the cause and the effect of the design decision represented by AR. Clause 2 of<br />
Definition 1 specifies that insertion of a link is not allowed if it would result in a directed<br />
cycle of links. Essentially, this means that the links maintain the integrity of the causal<br />
modelling whereby something cannot be the cause of itself, directly or indirectly.<br />
Using Definition 1, a basic <strong>for</strong>m of the model construct is {AE 1 , AE 2 , ...} → AR 1 →<br />
{AE a , AE b , ...} where AE 1 , AE 2 etc. are the inputs or the causes of a decision AR 1 ,<br />
and AE a , AE b etc. are the outcomes or the effects of the decision. Figure 6.2 shows a<br />
UML representation of the AREL model of the relationship between a motivational input<br />
AE, a decision AR and a resulting AE. The cardinality in the relationship shows that<br />
the motivational AE and the resulting AE must be a non-empty set linked by the single<br />
decision AR. The uniqueness constraint in the diagram specifies that each instance of AE<br />
in the relationship cannot appear more than once.<br />
Figure 6.2: A Causal Relationship between AEs and an AR<br />
The directional links of ≪ARtrace≫ represents the causal relationships in Figure 6.2.<br />
An AE causes AR by motivating or constraining the decision, and the AR results-in an<br />
outcome AE by having a design rationale to justify the design. Conversely, an outcome<br />
AE depends on AR which in turn depends on an input AE. The causal relationship is the<br />
basis <strong>for</strong> <strong>for</strong>ward tracing and the reverse is the implied dependency relationship which is<br />
the basis <strong>for</strong> backward tracing. Together they provide a means to navigate and explain the<br />
architecture design. This causal relationship simplifies and replaces similar relationships<br />
such as “creates” and “achieves”’ in argumentation-<strong>based</strong> methods (see Chapter 3). This<br />
is advantageous because the model becomes simpler and the complexity of traversal <strong>for</strong><br />
knowledge retrieval is reduced.<br />
85