Research activity in 1994
During 1994, research in the field of Knowledge Representation and Reasoning
has followed three main lines.
The first one, the most foundational one, has been devoted to studying the
representation of causal and temporal knowledge, with particular attention to
the representation of change, persistence and actions. Problems of commonsense
reasoning are strictly related to these issues, as well as nonmonotonic
reasoning forms. Significant results have been obtained in the use of
terminological languages for representing and reasoning with actionality, time
and causation.
A second field grew significantly during 1994: the one of temporal data bases.
Starting from past experience in temporal reasoning, this research activity has
been devoted to provide a foundation to temporal representation and reasoning
facilities that are being introduced in data bases. The problem of identifying
a good compromise between expressive power and computational complexity in
temporal database reasoning has also been considered, with significant results,
in particular regarding the complexity of propagation of quantitative and
qualitative temporal constraints.
The third line, which has been active for several years, is in the field of
model-based reasoning with particular attention to model-based diagnosis. In
1994 the attention has been devoted to modeling systems with time-varying
behavior and developing reasoning methodologies for diagnosis and monitoring of
such systems. Some results have been obtained augmenting a logical atemporal
model of the static properties of the system with transition graphs that model
possible changes in the behavior.
Since, even in case the system to be diagnosed is static, diagnostic reasoning
is complex, various opportunities for improving efficiency have been
investigated: compilation techniques have been used for focusing reasoning, and
a system has been experimented to integrated case-based reasoning with
model-based diagnosis.
Diagnostic reasoning has also been modeled through Petri Net analysis; such
methodologies provide insights on how parallel processing can be used for
diagnosis.
The research activity has been done in the Progetto Finalizzato CNR "Sistemi
Informatici e Calcolo Parallelo" (L.R.C. CAUSALI) (Computer Systems and
Parallel Computation, research line "Causal"), of the CNR Projects "Trattamento
di informazione temporali in basi di dati e basi di conoscenza" (Dealing with
temporal information in data and knowledge bases) and "Ambienti e strumenti per
la gestione di informazioni temporali" (Enviroments and tools for dealing with
temporal information), and of the bilateral France-Italy GALILEO project
"Model-Based Diagnosis".