Research activity in 1998
Model based reasoning has been the main area of investigation of our group in recent years. Several aspects have been explored.
A lot of attention has been given to the temporal dimension in model-based diagnosis, especially as regards the adequacy of different forms of temporal diagnosis for different classes of temporal diagnostic problems. In particular, approaches to modelling and diagnosis of dynamic systems have been analysed and compared.
Applications of model-based diagnosis have been developed in the BRITE-EURAM "Vehicle Model-Based Diagnosis" project. Particular attention has been given to the use of model-based diagnosis for on-board diagnosis of cars. Concrete results are the following:
A diagnosis engine has been developed; it can solve diagnostic problems as single cases, and it is also able to generate off-line a decision tree suitable to be embedded on board, to fit time and space constraints of an embedded system. The "Common Rail" system for diesel engine fuel supply has been modelled in terms of qualitative deviation equations, and the above diagnosis engine has been used for that model to produce a decision tree
As regards temporal reasoning in itself, we extended LaTeR, a manager of temporal constraints we have developed since 1993, in order to deal with constraints on intersection, union and difference of time intervals, and with disjunctive temporal constraints. We also applied LaTeR to different areas, ranging from temporal diagnosis to the treatment of temporal constraints in relational databases.
Moreover, we extended our previous work on periodic and repeated events in order to consider new forms of temporal constraints between periodic events, and to integrate such temporal constraints with the constraints on instances of events, such as those dealt with by the LaTeR system. We also start to investigate the possibility of extending standard relational databases to deal also with periodicity in both data and queries.
The research work in temporal reasoning has found application in the field of patient management and in temporal queries to database in industrial applications.
Related to temporal reasoning is the issue of reasoning about actions and change. In co-operation with the Catholic University of Leuven, a general solution to the ramification problem, that is, reasoning on side effects of events, has been developed.
Multi-modal reasoning
Since MBR is known to have high computational costs, the interest for combining MBR with other problem solving methods (potentially less expensive) has grown in the recent years. In particular we have investigated how to combine Case-based reasoning approach with MBR in order to speed-up diagnostic problem solving, and making it more practical also for large applications. During 1998 we have designed a major extension to the ADAPtER system, an architecture for diagnostic problem solving based on the integration of CBR and MBR.
By taking into account results obtained in 1997 in co-operation with University of Amsterdam, we investigated the reasons why multi-modal systems like ADAPtER suffer from utility problem (the increase of performance in one component does not implies an increase of performance of the whole system). We supplemented methodological investigation with an extensive set of experiments in order to single out the most relevant parameters influencing the utility problem. It turned out that the size and the content of the case memory were very relevant. For this reason research efforts were focussed in designing innovative retrieval mechanisms and, more important, learning algorithms able to decide automatically under which conditions new cases have to be learnt and stored in the case memory and when cases have to be forgotten (because non very useful for adaptation). Innovative results have been obtained in this area.
Finally, in the area of CBR, we have started a collaboration with the Medical Informatics Laboratory of the University of Pavia, with the goal of applying CBR techniques to medical problems. In particular, the suitability of particular retrieval strategies of past cases has been investigated in the field of diabetic patient management, with the aim of providing the physician with a suitable set of cases for insulin protocol revision.
In 1998 we have been involved in a project supported by ASI (Agenzia Spaziale Italiana) with other research centers and universities. The project is devoted to experiment innovative techniques for interactive autonomy with spacecraft. The domain chosen as test-bed is the space robotic arm SPIDER. In this co-operative project our group is responsible for diagnostic module. During 1998 significant amount of work has been devoted to study how to model the correct and the faulty behaviour of the robotic arm. In particular, we have studied how to make abstraction from the data and to model both local and global influence of the faults over observation.
1998 Publications
R. Bellazzi, S. Montani, L.Portinale. Retrieval in a Prototype Based Case Library: a Case Study in Diabetes Therapy Revision, in Proc. EWCBR 98, Lecture Notes in Artificial Intelligence 1488, pp. 64-75, Springer Verlag, 1998.
V. Brusoni, L. Console, B. Pernici, and P. Terenziani. Qualitative and quantitative temporal constraints and relational databases: Theory, architecture, and applications. IEEE Transactions on Knowledge and Data Engineering (to appear)
V. Brusoni, L. Console, P. Terenziani, and D. Theseider Dupre'. A spectrum of definitions for temporal model-based diagnosis. Artificial Intelligence Journal 102, 39-79,1998.
L. Console , R. Meo , D. Theseider Dupré: " On the automatic synthesis of functional dependency graphs from libraries of component models " in G. Rzevski, R.A. Adey, P. Nolan (eds.) Applications of Artificial Intelligence in Engineering XIII , pp. 591-608, Computational Mechanics 1998.
L. Console , D. Theseider Dupré: "On the dimensions of temporal model-based diagnosis", Proc. DX 98, 9th Int. Workshop on Principles of Diagnosis, pp. 16-23, 1998.
M. Denecker, D. Theseider Dupré, K. Van Belleghem: "An Inductive Definitions Approach to Ramifications" Linkoeping Electronic Articles in Computer and Information Science Vol. 3, pp. 1-38, (1998): nr 007
A. Guarnero, M. Marzuoli, G. Molino, P. Terenziani, M. Torchio, K. Vanni. Contextual and Temporal Clinical Guidelines. Journal of the American Medical Informatics Association, Symposium Supplement, AMIA Fall Symposium, pp. 683-687, 1998.
S. Montani, R. Bellazzi, L. Portinale, S. Fiocchi, M. Stefanelli. A CBR System for Diabetic Patient Theraphy, in Proc. Intelligent Data Analysis in Medicine and Pharmacology (IDAMAP 98), Brighton, J. Wiley & Sons Publ., pag. 64-70, 1998
A. Panati: "Diagnosi qualitativa di sistemi dinamici", AI*IA Notizie, Dic. 1998, p. 6-9.
L. Portinale, P.Torasso. Performance issues in ADAPtER: a combined CBR-MBR diagnostic architecture. Proc. AAAI Spring Symposium on Multi Modal Reasoning, pag. 47-52, AAAI Press, 1998.
L. Portinale, P.Torasso, P.Tavano. Dynamic case memory management. Proc. 13th Europ. Conf on Artificial Intelligence ECAI 98, H.Prade ed., pp. 73-77, Wiley, 1998.
M. W. vanSomeren, P. Torasso, J. Surma. A utility-based approach to learning in a mixed CBR-MBR architecture, in M.W. vanSomeren, H.J.P. Boshuizen, T. deJong, P. Reimann (eds.), Learning with Multiple Representations, pag. 283-304, Elsevier Science Publ., 1998.
P. Terenziani. Generating instantiations of contextual scenarios of periodic events. Proc. Fifth International Workshop on Temporal Representation and Reasoning - TIME98, pp. 61-68, IEEE Computer Society Press, Sanibel Island, Florida, 1998.
D. Theseider Dupré, A. Panati: "State-based vs. simulation-based diagnosis of dynamic systems", Proc. DX 98, 9th Int. Workshop on Principles of Diagnosis, pp. 40-46, 1998.
P.Torasso. Sulla rilevanza di artificiale in Intelligenza Artificiale. Sistemi Intelligenti, vol. X, n.3, pp.80-84, 1998
The research group was also active in the project devoted to intelligent agents co-ordinated at local level by Prof. Martelli and at national level Prof. Turini.
Moreover, the group co-operated with other members of the Department to the research activity carried on in the CANTIERE MULTIMEDIALE “Servizi telematici adattativi” sponsored by TELECM Italia.
