Information fusion and revision in qualitative and quantitative settings: steps towards a unified framework

摘要: Fusion and revision are two key topics in knowledge representation uncertainty theories. However, various formal axiomatisations of these notions were proposed inside specific settings, like logic, probability theory, possibility kappa functions, belief functions imprecise probability. For instance, the rule theory is Jeffrey's rule, characterized by axioms. The AGM axioms for stated propositional logic setting. But there no bridge between axiomatizations. Likewise, Dempster combination was axiomatized Smets among others, a logical syntax-independent axiomatization merging independently Koniezny Pino-Perez, while function can be viewed as weighted set. Moreover distinction fusion not always so clear comparing sets postulates each them enlightening. This paper presents tentative set basic principles another that could valid regardless whether information represented qualitatively or quantitatively. In short, obeys success postulate minimal change principle, essentially symmetric, principle optimism, tries to take advantage all sources information. Moreover, when pieces consistent, revising one other comes down symmetrically. Finally, commitment at work common operations.