Abstract:
The elicitation of requirements is a difficult and expensive
process but critical to the overall success of any system
development. So far relatively little work has been devoted to
providing intelligent tool support for this complex and labor intensive
activity. The quality of requirements from the elicitation
process currently depends greatly on the experience and expertise
of the participating requirements engineers, and the commitment
and cooperation of the system stakeholders.
In this paper we describe an agent-based approach to intelligent
tool support for requirements elicitation. Given the multiple roles
a requirements engineer must perform during elicitation, we
suggest a multi-agent system (MAS) may be developed as an
intelligent assistant for this process. It is proposed that some of
the tasks performed by requirements engineers during the
elicitation process may be supported and in some cases automated
by individual agents or several agents working cooperatively.
It is expected that the use of intelligent agents would produce
better requirements in terms of their completeness, correctness,
consistency and clarity. This would be achieved partly by
improving the elicitation process through greater efficiency with
respect to time and cost, and increased effectiveness by way of
rigorous and structured execution.