Abstract

The notion of context has been called to account for a wide range of linguistic and cognitive phenomena, some of which are relevant also from an AI perspective. However, the notion of context that emerges from the works in philosophy of language and logic seems to be inadequate for AI, because it presupposes some assumptions that are untenable from a methodological point of view. In the paper, I propose a change of perspective which is motivated by the work on context in AI. Instead of a way for structuring a description, either partial or complete, of the world, contexts are thought of as structured subsets of what an agent knows about the world. In the paper I argue that this is a promising basis for studying a logic of context. 1 Introduction The notion of context has been called to account for a wide range of linguistic and cognitive phenomena. A well-known example is that of indexical expressions, whose semantic value essentially depends on the (pragmatic) context of the...

Citations