Analysis of Early Aspects in Requirements Goal Models: A Concept-Driven Approach - Transactions on Aspect-Oriented Software Development - page 54

on “Security [C]” and one on “Confidentiality [D]”, and the other two rated four
on “Security [C]” and five on “Confidentiality [D]”. Thus, these two crosscutting
concerns are considered to be very similar based on grid analysis. If they are
indeed used interchangeably, correspondence would be established. Otherwise,
further elicitation should be performed to distinguish these constructs.
Although the customer and the developer both used the term “Usability”,
they probably did not refer to the same concept. These two constructs were
associated at the 80% level, one of the lowest matching scores between softgoals
appeared in Fig. 9. From the developer's perspective, the task “wildcard search”
affected “Usability” positively because people could do a fuzzy search and still
retrieve useful information. But the customer thought the task actually broke
“Usability” since using “wildcard” would involve a steep learning curve for a
non-technical user. Exploring such an inconsistency could spark the discussion
about what the concern “Usability” really meant to the media shop organization
and whether “wildcard search” should be implemented.
The grid analysis results, such as the correspondence and conflict relationships
identified above, not only enable us to gain insights into stakeholders' use of
terminologies and concepts, but also allow us to generate specific and plausible
hypotheses to be tested with subsequent efforts in eliciting and communicating
requirements. This indeed characterizes part of incremental analysis depicted in
Fig. 1, and leads us to assessing the RGT-based early aspects alignment method.
Assessment : Analogies can be drawn between RGT and structured interview.
Stakeholders undergo the interview by completing extracted and exchanged
grids. Analysis of resultant grids raises a plethora of new questions for further ex-
ploration. Essentially, a repertory grid is the start of a dialogue between analyst
and stakeholders. For example, if we have identified a potential correspondence
between two softgoals, we might ask the stakeholders to suggest further exam-
ples of concrete instances, to see if they confirm or refute the correspondence.
If stakeholders suspect their softgoals do not correspond, they may be inspired
to find tasks that disambiguate them. If they suspect two conflicting constructs
really do mean the same thing, they might discuss the apparent discrepancy in
their ratings of the concrete elements against this construct.
Assessment can thus be considered as a follow-up interview to address the
newly generated questions, and for stakeholders to provide evaluation and feed-
back about the quality and usefulness of the obtained data. In our framework,
there is no independent measurement since the collected data is context-laden
and is open to interpretation in a number of ways. Our exploratory RGT-based
approach is of practical value if our findings can provoke fruitful discussions and
guide further RE activities to precisely comprehend stakeholders' terminologies
and conceptual systems, thereby producing requirements that adequately reflect
their desires and needs.
Our concept alignment approach is appealing, since PCT and RGT avoids
the problems of imposition of terminology, and the meaning of a term is essen-
tially treated as a relationship between signs and actions. One desired outcome
of aligning concepts could be a vocabulary map between different viewpoints.