One of the benefits of formally represented knowledge lies in its potential to be shared. Ontologies have been proposed as the ultimate solution to problems in knowledge sharing. However even when an agreed correspondence between ontologies is reached that is not the end of the problems in knowledge sharing. In this paper we explore a number of realistic knowledge-sharing situations and their related problems for which ontologies fall short in providing a solution. For each situation we propose and analyse alternative solutions.

Data and knowledge bases model certain aspects of the world. When the state of the world changes faster than our ability to discover these state changes and update the data repositories accordingly, the confidence on the validity of data decays with time, and software systems running in such environments have to cope with the decay of con dence in the data lest they run the risk of giving wrong answers and behaving erroneously. This gradual loss of confidence on stored data is termed information obsolescence, and it is inherently a temporal phenomenon. We have come across such problem when designing an information system for traffic monitoring and control in a large city, and in this paper we propose to investigate some problems related to this modelling task. We propose two approaches to deal with information obsolescence, coined here the analytical approach and the algebraic approach, and show how both can converge to a general, temporal treatment of obsolescence. Our immediate goal ...

. In this paper we compare two well-known techniques for reasoning with uncertainty --namely, Incidence Calculus and the Theory of Evidence -- from a viewpoint not so frequently explored for such techniques. We argue that, despite the equivalence relations that these techniques have been proved to hold, they have intrinsically different roles as representations of uncertainty for automated reasoning, in the sense that the former represents approximations to uncertainty values due to impossibility to achieve exact results by proof-theoretic means, and the latter represents model-theoretic limits of definability of uncertainty values. Keywords: representation languages for uncertain knowledge. 1 Introduction As it has been frequently argued for, reasoning with uncertainty is a "kaleidoscopical " task, since uncertainty itself is a multifaceted concept [Cla90, Coh86, CdSRH93, Saf87]. Not surprisingly, different facets of uncertainty and different applications and domains for automated ...

. Uncertainty is a multifaceted concept, and a system for automated reasoning with multiple representations of uncertainty was proposed in [CdSRH93]. In this work we base on that system and present an efficient language for reasoning with fuzzy predicates. The language in [CdSRH93] is developed as a PROLOG meta-interpreter. Since the first implementation of this language is not very efficient, we propose two optimization strategies to improve its computational efficiency in time. In order to avoid redundant or unnecessary intermediate computations, we employ two classical optimization techniques, respectively solution caching and ff-fi pruning. Keywords: uncertainty, uncertain reasoning, fuzzy sets, fuzzy predicates, fuzzy reasoning. 1 Introduction Uncertain Reasoning - the general denomination given to the problems of reasoning with and about uncertainty - is an interesting and challenging issue for researchers in Artificial Intelligence. One of the multiple forms of uncertainty t...

This paper contains some methodological reflections upon developing good-quality research in Artificial Intelligence. We consider in these reflections research programmes that aim at achieving basic scientific results together with developing techniques for use in applied systems. We focus on an empirical approach to research organisation, in which the results can be assessed through the technical robustness of the scientific theories developed, as well as through the success of applications that are built using the obtained techniques. We discuss this approach through an example, namely we present the development of an Intelligent Tutoring System, and show some issues that can arise when such approach is employed. 1 Introduction Artificial Intelligence is still a new area of research, and one clear indication of this is the fact that the precise definition of what it is about is still rather contentious. Our personal view about how Artificial Intelligence can be defined begins with...

Knowledge sharing and reuse has been considered an important issue for cost-eective use of knowledge-based systems, especially after the development and popularisation of objectbased technologies and Internet-based decentralised computing. Up until now, the majority of research tackling this issue has been founded on the assumption that there can be a common domain description { a shared ontology { which suits everyone with an interest in the knowledge. Unfortunately, getting an agreed ontology for a collection of systems can be a dicult problem and, even when this problem can be solved, it may not be enough for eective knowledge sharing, since the way we represent knowledge is intimately linked to the inferences we expect to perform with it. A nice example of this situation can be found in systems for reasoning under uncertainty, where even if we do have a shared ontology for the problem being solved we must still establish semantic links between the inferences performed within each system to actually have knowledge being shared and reused.