|
The goal of the decision management applications
is to develop a tool that better prepares people in an emergency
situation to react swiftly and efficiently to a disaster.
For attaining this goal, the semantic level information is
a key to addressing the problem of information overload on
the crisis managers. And there are three kinds of knowledge
which must be represented in this domain; (1) Crisis scene
and human behavioral data analyzed for higher-level concepts,
features, trends, patterns, etc., (2) Disaster/crisis scenario
domain knowledge, and (3) Knowledge about crisis workers/managers.
With above knowledge, crisis worker/manager request can be
used to generate interactive presentations back to the user
to assist them in their tasks. We customize the information
presentations they are provided in response to requests or
events. Our approach is to construct a federation of ontologies
to handle the semantic meta-data for each kind of knowledge
illustrated in figure 1.

Figure 1. Crisis Manager “Portal”
Our main role in this project has two key aspects:
(1) Analysis and mining event history archive: Using domain
ontology, we generate a history archive of events that take
place within a given geo-spatial time frame. And monitor decisions
and actions taken during an event and review these decisions
for future scenarios. And then learn from the previous “similar”
encounters. At last, mine histories for novel/problematical
patterns using task-domain ontology.
(2) Filtered interactive multimedia presentation with user
model: The system is connected to the 3D visualization, which
enables the fire commander to assign fire engines to specific
fires. And there are more multi-modal information about the
fires such as text, image, and so on. The multimedia data
presentation will be displayed to the crisis manager in the
way allows them to quickly recognize the crisis situation
and problem(s). And the presentation will be customized using
the user model.
|