Expression of uncertainty in linguistic data

This paper briefly introduces several of the aspects to take into account in order to properly describe and analyze the expression of uncertainty in textual data. Different types of ambiguity inherent to the nature of language itself are presented. Linguistic ambiguities can be observed between symbols and the meanings arbitrarily attached to them. Many natural language processing techniques can be applied to texts to minimize linguistic ambiguities. Referential ambiguities relate to the world and can be observed through extra-linguistic environments, each potentially impacting the interpretation of natural language utterance. From a linguistic point of view, the identification and automatic tagging of expressions of certainty/uncertainty in textual data is a sine qua non condition to enable the empirical study and modeling of how humans assess certainty through their use of language. Such analysis is required to generate future language-dependent models of certainty/uncertainty suitable for information fusion systems.