China
Africa
Explore chapters and articles related to this topic
Agent Architectures
Published in Weiming Shen, Douglas H. Norrie, Jean-Paul A. Barthès, Multi-Agent Systems for Concurrent Intelligent Design and Manufacturing, 2019
Weiming Shen, Douglas H. Norrie, Jean-Paul A. Barthès
Having knowledge is not sufficient: an agent needs to store and process such knowledge locally. Services such as content-based information distribution1 impose requirements on how local knowledge is to be processed. For example, deductive inference is needed for content-based distribution, truth-maintenance capabilities are needed for belief revision, and so on. Other requirements may come from the enterprise modeling domain: the ability to represent complex enterprise models or to reason about common-sense notions like time. The following mechanisms may be needed in the knowledge management module: - Terminology construction and verification: defining the terms of a domain by logical composition mechanisms, automatically organizing terms in taxonomies, ensuring consistency of definitions (see Chapter 11 for details).- Model construction: building actual models of domains using the defined terminologies; ensuring consistency of these models.- Multi-agent belief revision: supporting belief revision among multiple agents in a manner preserving consistency among the participating agents.- Common sense representation and reasoning about time.- Model change management: monitoring the dynamic modification of models by notifying interested agents when changes in their areas of interest occur.
Abduction in argumentation frameworks
Published in Journal of Applied Non-Classical Logics, 2018
Boella et al. (2009a, 2009b) consider the effect of adding/removing arguments or attack relations under the grounded semantics. Cayrol et al. (2010) study the effect of an addition of an argument on the outcome of the argumentation semantics. The goal of these studies (Boella et al., 2009a, 2009b; Cayrol et al., 2010) is identifying possible changes of extensions after revising an argumentation framework, which is in contrast with our goal of identifying possible changes of an AF to have a particular outcome. Falappa et al. (2002) introduce a belief revision framework in a structured argumentation system where K represents undefeasible beliefs and Δ represents defeasible beliefs. Given a sentence α, a set (called an argument) is computed such that and where ⊢ represents defeasible inference. The set T is then incorporated into K to obtain a revised knowledge base. The study computes explanations for revising an argumentative framework, while it considers a structured argumentation system and is different from the abstract argumentation framework considered in this paper. Since there is a close connection between structured argumentation and defeasible logic (Lam, Governatori, & Riveret, 2016), abduction in (structured) argumentation would be related to revision in defeasible logic.
Morphologic for knowledge dynamics: revision, fusion and abduction
Published in Journal of Applied Non-Classical Logics, 2023
Isabelle Bloch, Jérôme Lang, Ramón Pino Pérez, Carlos Uzcátegui
We start with some basics about belief revision. The aim of belief revision is to model how to incorporate coherently a piece of information to a corpus of beliefs. In the most studied model, the AGM model (Alchourrón et al., 1985), the corpus of beliefs is represented by a logical (consistent) theory K and the (new) piece of information by a formula ψ. The result of incorporating ψ to K, i.e. the revision of K by ψ, is denoted by . We give here a very simple presentation of this model in finite propositional logic due to Katsuno and Mendelzon (1991) in which the (old) beliefs K are indeed represented by a consistent formula φ (that is, ) and the revision of φ by ψ is denoted . Note that ° is a function mapping an ordered pair of formulas into a formula. This kind of function is called a revision operator5 when it satisfies the following rationality postulates: (Success)If then (Minimality)If then (Coherence)If and then (Syntax independence) (Superexpansion)If then (Subexpansion).