Connectionist classification and specific stochastic models in the understanding process of a dialogue system

摘要

In this paper we present an approach to the application of specific models to the understanding process of a dialogue system. The previous classification process is done by means of Multilayer Perceptrons, and Hidden Markov Models are used for the semantic modeling. The task consists of answering telephone queries about train timetables, prices and services for long distance trains in Spanish. A comparison between a global understanding model and the specific models is presented.