Agent concepts have been used in a number of recent approaches of requirement engineering (RE),such as KAOS (Knowledge acquisition in automated specification), i* and GBRAM (Goal-based requirements analysis method). And the modeling languages used in those approaches only permit precise and unambiguous modeling of system properties and behavior. However, some system problems, particularly those drawn from the agentoriented problem domain, may be difficult to model in crisp or precise terms. There are several reasons for this. On one hand, the lack of information may produce the uncertainty of the class to which an object belongs. If we have enough information or if we are considering sufficient attributes,we should be able to make a precise categorization. On the other hand, uncertainty may also arise from some natural imprecision in requirement describing itself, such as soft goal describing and uncertain concepts describing. In the second case, the classification into precise classes may be impossible, not because we do not have enough information, but because the classes themselves are not naturally discrete. In this paper, we start with a discussion of the uncertainty in agent-oriented requirement engineering. Then we propose to handle the uncertainty using fuzzy sets. Finally we refine this proposal to integrate a fuzzy version of Z with the KAOS method. This integration is illustrated on the example of the mine pump. In the conclusion part,we compare the advantages of our approach with those of the classical KAOS approach.
展开▼
