This paper describes how robust parsing techniques can be fruitful appliedfor building a query generation module which is part of a pipelined NLParchitecture aimed at process natural language queries in a restricted domain.We want to show that semantic robustness represents a key issue in those NLPsystems where it is more likely to have partial and ill-formed utterances dueto various factors (e.g. noisy environments, low quality of speech recognitionmodules, etc...) and where it is necessary to succeed, even if partially, inextracting some meaningful information.
展开▼