<?xml version="1.0" standalone="yes"?>
<Paper uid="A94-1017">
  <Title>Real-Time Spoken Language Translation Using Associative Processors</Title>
  <Section position="6" start_page="105" end_page="105" type="concl">
    <SectionTitle>
5 Conclusion
</SectionTitle>
    <Paragraph position="0"> This paper has proposed TDMT (Transfer-Driven Machine Translation) on APs (Associative Processors) for real-time spoken language translation. In TDMT, a sentence is translated by combining pieces of transfer knowledge that are associated with examples, i.e., source word sequences. We showed that the ER (example-retrieval) for source expressions including a frequent word, such as a function word, are predominant and are drastically speeded up using APs. That the TDMT using APs is scalable against vocabulary size has also been confirmed by extrapolation, i.e., a 10-AP sustained performance to an 800-AP expected performance, through analysis on communications between APs. Consequently, the TDMT can achieve real-time performance even with a large-vocabulary system. In addition, as our previous papers have shown, the TDMT achieves accurate structural disambiguation and target word selection. Thus, our model, TDMT on APs, meets the vital requirements for real-time spoken language translation.</Paragraph>
  </Section>
class="xml-element"></Paper>