Information Technology Reference
In-Depth Information
their perspectives. Our approach takes the view based on personal constructs
theory, which offers a mechanism to compare and contrast objects in the do-
main of interest. Our PCT- and RGT-based concepts alignment method avoids
the problems of imposition of terminology when individuals construe and de-
scribe concepts in the problem domain, and the meaning of a term is essentially
treated as a relationship between signs and actions in our framework. NFR or
early aspects catalogue building could benefit from our approach since a well-
categorized taxonomy is viewed as a catalogue of how constructs represented by
linguistic symbols relate formally in a particular context.
Natural language processing (NLP) plays an important role in identifying as-
pects in requirements. Many early aspects frameworks (e.g. [6, 41, 46]) adapt
certain steps in linguistic engineering (flatten, tokenize, stem, and remove stop
words) so that aspect words (e.g. verbs that scattered in multiple requirements)
can be extracted from requirements statements. Although NLP-based techniques
could reach high recall and precision under certain circumstances, taking for
granted that natural language-based requirements statements are an unprob-
lematic starting point is a historical accident rather than a position grounded in
the realities for RE [47]. We assume that there exists a relatively well-organized
set of requirements models, and present a concept-driven framework that takes
advantage of these models. Our approach complements existing NLP-based early
aspects methods by providing mechanisms to capture and analyze overlapping,
corresponding, conflicting, and crosscutting concerns addressed in fine-grained
requirements models.
FCA has typically been applied in the field of software engineering to support
software maintenance activities [49, 51], such as program understanding [52],
object-oriented class identification [50], reengineering of class hierarchies [44],
dynamic analysis [9], and software configurations [20]. In the analysis of soft-
ware systems, especially source code exposing certain structural and behavioral
properties, several relationships among the composing entities emerge. For this
reason, FCA has found a very productive application area associated with static
and dynamic analyses. Recent work, such as [36], has also reported the ap-
plication of FCA in RE activities. The proposed approach exploits structural
properties in requirements goal models, in particular, the binary contribution
relations between tasks and softgoals in those models. Our novel application en-
hances the overall competence of FCA in exploring crosscutting properties in
the early phases of software development.
The idea of translating RGT into FCA due to their common cross-reference
data structure is not new. Much work has been carried out in the knowledge
engineering field. For example, DelugachandLampkin[7]presentedahybridap-
proach containing RGT and FCA to facilitate the knowledge acquisition process
of transferring and transforming information from the domain experts to the
expert system. In their approach, the traditional “triad” method [10] associ-
ated with RGT was applied to elicit knowledge from domain experts, and the
translation from RGT to FCA was performed via an intermediate knowledge rep-
resentation schema — conceptual graphs [45]. In contrast, our approach outlines
Search WWH ::
Custom Search