<?xml version="1.0" standalone="yes"?>
<Paper uid="N03-3010">
  <Title>Cooperative Model Based Language Understanding in Dialogue</Title>
  <Section position="6" start_page="0" end_page="0" type="concl">
    <SectionTitle>
5 Conclusions
</SectionTitle>
    <Paragraph position="0"> In this paper we proposed a cooperative model incorporating finite state model and statistical model for language understanding. It takes all of their advantages and suppresses their shortcomings. The successful incorporation of the methods can make our system very robust and scalable for future use.</Paragraph>
    <Paragraph position="1"> We notice that the series model of the finite state machine model actually incorporates some semantic knowledge from human beings. Ongoing research work includes finding new ways to integrate semantic knowledge to our system. For the statistical learning model, the quality and the different configurations of training set highly affect the performance of models trained and thus their abilities to process sentences. The balance of training set is also a big issue. How to build a balanced training set with single finite state machine model will remain our important work in the future. For the learning mechanism, Naive Bayesian learning requires more understanding of different factors' roles and their importance. These problems should be investigated in future work.</Paragraph>
  </Section>
class="xml-element"></Paper>