Semantic Requirements Engineering - Intentional Perspectives on Information Systems Engineering - page 75

Figure 7 shows the process for the tool user to add a missing goal to a goal
graph. After loading the ontology of reservation business domain and then starting
inference, the tool detects the candidates of missing goals. The mark of “boxed +”
symbol that is attached to a goal shows that there may be missing goals related to
it, and the tool user can refer to the candidates of missing goals by clicking the
mark with a mouse. In Fig. 7, our tool tells the user that goal N30 “Reservation by
users themselves” may have missing goals because the mark of “boxed +” symbol
appears in its upper left area. Clicking the mark allows her or him to see a list of
the candidates of missing goals, and as a result three candidates are shown. The
ontology used in this example contains the concepts “reservation”, “cancel” and a
relationship “requires” between them, as shown in Fig. 6. This part of the ontology
denotes that “reservation” requires “cancel”. Since the current version of the graph
does not include a goal related to the “cancel” concept yet, the tool suggests the
addition of the goal having “cancel”. In addition, the other concepts “log in” and
“reservation information” are suggested as missing goals. The tool user accepts a
candidate of a missing goal “cancel”, and then she or he adds a new goal at the
suggestion of the dialog shown in Fig. 7. The dialog enables her or him to fill the
appropriate name of a new goal, and to connect the new goal to the existing goal(s).
In this case, the new goal N31 is named “Cancellation enabled”, and the goal is
connected to the goal N30 as a sub-goal. The technical details such as the structure
of domain ontologies and inference rules for deriving missing goals are discussed
in [18] .

4 Semantic Quality Metrics

In IEEE 830 standard [5] , there are eight characteristics such as correctness and
completeness to measure the quality of software requirements specification doc-
uments. Although this standard includes several methods to measure the quality
characteristics, most of them are related to syntactical aspects of a specification
document. For example, the IEEE 830 standard says that we should check whether
all figures, tables, and diagrams in the document are labeled and referred, in order
to measure its completeness. In the usual sense, completeness denotes no miss-
ing requirements in the document and it should include the semantic aspects of
the documents, i.e. there does not exist semantically lacking of requirements in the
document. In the example of Fig. 1 in Sect. 1, “stop” message sending is missing
in the diagram and it is semantically incomplete. By using the semantic mapping
of requirements to ontological elements, we can estimate the degree of seman-
tic incompleteness. More concretely, we can calculate the ratio of the ontological
elements required but not mapped from requirements. In Fig. 3, we have 5 ontolog-
ical elements mapped from the sequence diagram and one missing element “stop”.
Thus semantic incompleteness of this diagram can be estimated as 1/5 20%. We

can define the other quality characteristics based on semantic aspects using our
ontological approach, and readers can find their details in [ 6] .

=