The approach used for tests in ASPECS consists in successively testing each context from the role (the smaller one) to the entire system. This activity aims at testing behaviors that will be used to compose the system; this means individual behaviors represented by roles and global behaviors corresponding to organizations.
Each role is individually tested; stimuli coming from other roles defined within
the same organization and stimuli coming from the agent context through capacities
are emulated (here the referred agent is the one that should play the role under
test). This kind of test involves the definition of stubs and drivers like it happens
in conventional object-oriented unit test. Each organization is then instantiated in
a runtime context called group. Its roles are instantiated too. This task consists in
validating roles interaction scenarios inside an organization, stimulus coming from
agent contexts are emulated. From the input artifacts, it is possible to produce a set
of test cases that can verify the obedience of the obtained roles and organizations to
the prescribed specifications. After writing test cases and coding necessary support
elements, tests can be performed and identified failures will lead to new iterations
to fix related bugs.
The ASPECS process also enables the use of formal methods, such as model checking and theorem proving [1].
References
- [1] A Verification by Abstraction Framework for Organizational Multi-Agent Systems
- N. Gaud, V. Hilaire, S. Galland, A. Koukam, and M. Cossentino.
In AT2AI-6@AAMAS’08, Estoril, Portugal, May, 2008.
