Information Technology Reference
In-Depth Information
In John's case study, when John uses the system, the physical context is 10:00AM,
Sept 3
rd
, 2012, in the Olympic Village in London according to the sensors. Based on
John's i*-Context (Table 2), his foreground social position is
Position
[
Father
]. This
is because he has already finished his race, and now doing something to
compensatefor being away from his little daughter. According to his to-do list as
Position
[
Father
], buying a gift for his daughter is on the top of the list.
Step 2. Creating Goal-Task Decomposition Tree:
The second step of the process focuses on the user's needs, and uses iterate
refinement technique to create a goal-task decomposition tree based on the top user's
goal.
In John's case study,
Goal
[
Buy Gift for His Kid
] could be satisfied either online or
at a gift shop. When we look into
Task
[
Shop at Stores
], the task could then be further
decomposed into three separate sub-goals, namely,
Goal
[
Choose Stores
],
Goal
[
Find
Traffic Route
] and
Goal
[
Choose Gift
]. After that, these sub-goals are being
operationalized by a set of different web services (modeled as the bottom-level yellow
hexagons). In this step, softgoals derived from goal and task decomposition will also
be modeled (see
Softgoal
[
Gift to be Appropriate
]).
Step 3. Generating Preference Softgoal Structure:
In the third step of the process, preference softgoals are being deduced from user's
agent classes, his foreground position and the physical context.
As shown in Fig. 7, according to the pre-defined preference deduction rules, John's
preference softgoal structure can be generated. For instance,
Softgoal
[
Wheel Chair
Accessible
] is introduced since hebelongs to
Agent Class
[
wheel chair user
]. While as
Softgoal
[
Gift to be Exotic
] is derived from the existing
Softgoal
[
Gift to be
Appropriate
], because John is from China and he is currently in U.K. Two other
preference softgoals can be acquired in the same way.
Step 4. Selecting & Customizing Services:
In the last step of the process, the alternative services obtained from step 2 will be
further selected and customized to better comply with the user's preference softgoals.
As the bottom-level service tasks and the preference softgoal being identified, the
relationships between these services and softgoals could be used for service selection
and customization. Assuming all the web services are stateless services, the
relationships can be represented with the existing conditions of the contribution links
in between. Take relationship number 3 as an example,
Service
[
Public
Transportation Planning
] could make the
Softgoal
[
Wheel Chair Accessible
] possible,
only if the information service provide a barrier-free access route. Thefinal service
selection and customization result after this process is shown in the lower left corner
of Fig. 7.
5 Related Work
We discuss related work in three areas: agent intentional modelling in requirements
engineering, context-aware web services research and preference-based service
selection and customization practices.
