One of the problems of the i* approach is the lack of explicit process
guidance about what to do in what order. I have sought to overcome
this by providing concrete guidance, especially when developing the SR
model. There are 3 key stages to undertake.
The first stage is to develop each basic SR model by modelling what
each actor can accomplish in its own, without depending on other
actors. One such model is produced for each actor.
The second stage begins to link these independent SR models together
using the dependencies specified in the SD model. This is quite
straightforward – we add the dependency links between actors in the
SD model to the SR elements of those actors in the SR model.
The third stage completes this integration process. We use the
dependencies between the actor SR models to integrate them into one
system-wide SR model.
But where to start? There are 2 basic starting points for these models. If
a specific solution has been mandated, model it as a task and
decompose it, as well as ask why to discover the goals and soft goals.
If not such solution has been mandated, start from the most important
goals and soft goals, and develop goal hierarchies and contribute-to
hierarchies. You can check the completeness of these independent SR
models using the SD model
Reference: Neil Maiden (2011) Requirements Engineering Lecture Notes