Information Technology Reference
In-Depth Information
bring in new ones if a particular event occurs, etc., without having to configure
anything. The presentation devices must be capable of adapting themselves to
change without a human intervention being necessary. This is why decentralized
software architecture based on the notion of an agent is required in this instance.
Each of the three types of entities mentioned previously corresponds to a type of
agent:
- user agent (U): acts as active software representation of human users;
- information agent (K): corresponds to the software declination of information
sources - it provides information to the user agent;
- presenter agent (P): constitutes a software interface with the physical
presentation devices.
Thus, the world of agents constitutes a “mirror” of the real world regarding our
three entities of interest. We presume that all agents can communicate with each
other. Communication can be via wireless networks such as Wi-Fi. The
relationships of sensory proximity in the physical world are mirrored in the world of
agents. For example, if user
a
perceives a presentation device
b
,
then the same
relationship will exist between the associated agents. Agents are reactive: they are
dormant most of the time and react when particular events occur. In practice, a
given agent
a
can react to three kinds of events:
- another agent
b
comes towards
a
3
;
- agent
b
, who was close to
a
, has just moved away from it;
-
a
has just received a message via the network, coming from a random agent
c
,
which is not necessarily close to
a
.
Thus, if the agents were to find themselves on their own in the system, nothing
would ever happen. Agents have reactive behaviors when the physical entities that
embody them move. This means that the proactivity of the system is ensured by
physical entities, in particular human entities: it is the latter that will generally move
around, and from there set off a cascade of reactions in the system.
10.6. Allocation and instantiation in KUP
Allocation consists of selecting the most adequate presentation modality for a
given semantic unit (depending on the user, the device and the information to be
presented). As for instantiation, it concerns the definition of the most adequate
values for the attributes of the selected modality [ROU 06]. In KUP, the allocation
3 In the sense of sensory proximity.
