In order for artificial intelligence to become truly
useful in real-world applications and environments it is necessary to identify,
document, and integrate into automated systems the human knowledge that people
use to solve their real-world problems. This process has been found to be
difficult, and is a critical part of what has become known as the knowledge
acquisition bottleneck. The primary
contribution of this dissertation is the development and application of a
general methodology for modeling and representing an expert’s problem-solving
knowledge that supports ontology import and development, teaching-based
intelligent agent development, and agent-based problem solving.
The methodology provides practical guidance to subject matter experts on
how to express the way they solve problems using the task reduction paradigm.
It identifies the necessary concepts and features to be represented in
the ontology; identifies the tasks to be represented in the agent’s knowledge
base; guides the rule learning and refinement processes; supports natural
language generation of solutions and justifications, and is natural and easy to
use. The methodology is applicable
to a wide variety of domains and has been successfully used in military
planning, course of action critiquing, and strategic center of gravity
identification problems. This
research is part of a much larger effort with the goal to develop an advanced
approach to expert knowledge acquisition based on apprenticeship multi-strategy
learning, in a mixed-initiative framework.
|