Requirements as Goals and Commitments Too - Intentional Perspectives on Information Systems Engineering - page 149

with others. Our proposal differs in that cooperation is guaranteed by mutual interest
in a commitment: the agents playing debtor and creditor have their own reasons to
interact via commitments, but they don't (and can't) know the other party's motiva-
tions. Penserini et al. [ 25] have extended Tropos to design web services that support
the stakeholders' goals. The main limitation of this approach is that it assumes that
requirements engineers have a global view on all the actors.
KAOS [ 10] exploits a system-oriented perspective to specify requirements.
Stakeholders are essential to gather system goals, but they are not explicitly rep-
resented in KAOS models. Leaf level goals are assigned to agents on the basis of
a responsibility principle; van Lamsweerde has also discussed how KAOS require-
ments models can be mapped to software architecture [ 40] . KAOS is effective for
the development of traditional software systems, but lacks of the proper abstractions
to design autonomous and heterogeneous agents in open systems.
Gordijn et al. [ 19] combine i ∗ goal modeling with profitability modeling for
the various stakeholders to design e-services. In such a way, the authors consider
not only the intentions of the agents, but also the economic value of a service.
Their approach is less generic than ours: economic value exchanges are a very
important criteria but not the only one; moreover, they assume a monolithic
system-development point of view which does not suit well in open systems.
Liu et al. [22] propose an i ∗ extension intended for the design of open systems,
and propose some reasoning techniques that can be executed against these models.
The authors formalize commitments in a weaker sense—as a relation between an
actor and a service, not between actors, as is done in our approach.
Bryl et al. [ 5] use a planning-based approach to design socio-technical systems.
The main intuition behind this work is to explore the space of possible alternatives
for satisfying some goal. However, unlike us, they follow goal dependencies inside
the dependee actors, thus violating heterogeneity.
6 Conclusion
The power of any technique for eliciting, modeling and analyzing requirements rests
on the primitive concepts used to conceptualize them. The advent of goal-orientation
in RE twenty years ago brought about a shift from a functional to an intentional view
of software systems. The implications of this shift are still being worked out.
This chapter advocates a further shift from an intentional to a social view of
requirements for socio-technical systems. The proposal continues along a path orig-
inally defined by i ∗ in Eric Yu's PhD thesis. Our new proposal is founded on the
concept of commitment and related social concepts; it calls for a new form of sys-
tem specification that prescribes a system's course of action more concretely than
goal-oriented techniques, but more abstractly than process-oriented ones. We see
this proposal as yet another step towards an agent-oriented view of socio-technical
systems, their conceptualization, design, and evolution.