Donald Bren School of Information and Computer Sciences

The term "inference" has been used recently in computational linguistics and artificial intelligence to refer to any conjecture or conclusion drawn from a text. Presupposition and entailment are a subclass of inferences that appears to be tied to the structure of language, for they arise from the semantics of particular words and from syntactic constructs of the language. As a subclass of inferences, presupposition and entailment exhibit several properties that do not hold for the general class of inferences; we examine some of those properties here. In particular, we show how to compute the presuppositions and entailments of a sentence while parsing. The computation is by structural means (e.g., uses tree transformations) using an augmented transition network.