Abstract

In most natural language processing applications, Description Logics have been used to encode in a knowledge base some syntactic, semantic, and pragmatic elements needed to drive the semantic interpretation and the natural language generation processes. More recently, Description Logics have been used to fully characterize the semantic issues involved in the interpretation phase. In this chapter the various proposals that have appeared in the literature about the use of Description Logics for natural language processing will be analyzed.

Introduction

Since the early days of the Kl-One system, one of the main applications of Description Logics has been for semantic interpretation in natural language processing [Brachman et al., 1979]. Semantic interpretation is the derivation process from the syntactic analysis of an utterance to its logical form – intended here as the representation of its literal deep and context-dependent meaning. Typically, Description Logics have been used to encode in a knowledge base both syntactic and semantic elements needed to drive the semantic interpretation process. One part of the knowledge base constitutes the lexical semantics knowledge, relating words and their syntactic properties to concept structures, while the other part describes the contextual and domain knowledge, giving a deep meaning to concepts. By developing this idea further, a considerable part of the research effort has been devoted to the development of linguistically motivated ontologies, i.e., large knowledge bases where both concepts closely related to lexemes and domain concepts coexist.