Semantic Requirements Engineering - Intentional Perspectives on Information Systems Engineering - page 77

Artifact Delta = {
1. Add the use case Deposit.
2. Add the association between Customer and Deposit.
3. Add "or an exchange button" at line 4 after the 4th word.}

Withdraw
1. Insert a bank card into an ATM.
2. Input a PIN code.
3. Input an amount.
4. Touch a confirmation button.
5. Get the bank card and cash.

Withdraw

Customer

Customer

Deposit

1. Insert a bank card into an ATM.
2. Input a PIN code.
3. Input an amount.
4. Touch a confirmation button
or an exchange button.
5. Get the bank card and cash.

Ver. 1

Ver.2

Ontology for Ver.1 =
{Customer, Withdraw, Bank card, ATM, PIN,
Amount, Confirmation, Cash}

Ontology for Ver.2 =
{Customer, Withdraw, Deposit , Bank card,
ATM, PIN, Amount, Confirmation, Cash,
Exchange }

Ontology Delta = {1. Add Deposit. 2. Add Exchange.}

Fig. 9 An example of semantic version control

respectively. The artifact delta d j (0 n -1) is produced as the difference between

V j and V j +1, while Od j is the difference of ontological elements between Od j and
Od j +1, i.e. ontology delta. These ontology deltas allow us to capture the changes of
artifacts semantically and reasoning about semantic aspects of their changes can be
automated.
Figure 9 shows the simple example. Suppose that we model a banking business
and specify it as a use case model. The left hand of the figure depicts a use case dia-
gram having only one use case Withdraw and its use case description with natural
language sentences. A requirements analyst adds the use case Deposit and the func-
tion of exchange during Withdraw in a newer version Ver.2. As a result, the ontology
delta consists of the addition of two ontological concepts Deposit and Exchange.
By tracing the ontology deltas, the analyst can find what functions had been
semantically added, deleted or modified in which versions of the use case models.

≤ j ≤

6 Conclusion

This chapter presents the idea of using ontologies as a semantic basis in require-
ments engineering (RE) so as to automate semantic processing in RE techniques.
As mentioned in each section, this idea is not new and it already appears in
specific applications such as the support of missing requirements sentences [ 7]
and of suggesting sub-goals in GORA [18] , semantics of meta models [2, 4] and a