After a problem solving method and
a domain ontology has been identified, there remains still the problem
of adding sufficient and appropriate domain knowledge to the system.
This track will cover support tools
for the Knowledge Acquisition process - up to the extent that the expert
directly interacts with the machine - without any Knowledge Engineer.
That is, a generic problem-solving method would be used, which is turned
into a pdomain-specific method by parametrization through a direct system-expert
interaction. Similarly, the domain ontology would be designed by direct
system-expert interaction. In this case, the expert would not be involved
in designing the problem solver - rather the system designs the problem
solver based on its interactions with the expert.
Conceptual ideas and analysis of
possibilities and limitations are sought as well as actual systems. The
range of ideas and systems include fully-automated expert interviews, Machine
Learning techniques to propose additions or changes to a Knowledge Base
as well as approaches which allow an incremental approach to designing
a knowledge base during use of the Knowledge-Based System. That is,
facilities for allowing the expert to patch the knowledge base locally.
Problems involve the question of what sort of additions and changes to
a knowledge base can be safely controlled by an expert.