|
In this large and complex area different research activities converge with the
aim of defining powerful models for knowledge representation, suitable
for wide spectrum applications and domains. Starting
from standard logic programming, extensions have been defined to
support hybrid models, integrate hypothetical reasoning,
viewpoints, inheritance and abductive reasoning. Jointly
with IBM Watson Research Centre, a new support model for hypothetical
and qualitative reasoning has been defined, which also provides a
distributed
implementation. This model has been used to analyze complex physical
domains.
In the area of hybrid knowledge representation systems the Entity-Situation
model has been proposed. It synthetizes ideas from the E-R model, developed
in database environment and the KL-ONE model, developed in AI
environment. Its main features are the representation of n-ary relationships
between concepts and
the availability of a sound, complete and tractable subsumption algorithm.
A hybrid knowledge representation system based on the
Entity-Situation model has been implemented in prolog. It manages
terminological knowledge and recognizes the assertions with respect to a
given terminology.
|