<?xml version="1.0" standalone="yes"?>
<Paper uid="C82-1044">
  <Title>AN ENGLISH-JAPANESE MACHINE TRANSLATION SYSTEM BASED ON FORMAL SEMANTICS OF NATURAL LANGUAGE</Title>
  <Section position="7" start_page="277" end_page="277" type="evalu">
    <SectionTitle>
EXPERIMENTS
</SectionTitle>
    <Paragraph position="0"> A prototype of the system has been constructed on a personal LISP system (Voshita (1978)), which is developed on a minicomputer with LISP-oriented storage subsystem. As to the analysis module, sixth version is in use; as to the generation module, first version is in use. About two years since the last COLING conference at Tokyo were mainly devoted to the development.</Paragraph>
    <Paragraph position="1"> At the first stage of experiment, sample sentences were tested for several sentence patterns. At the second stage, our purpose was to extend the system for practical test; to translate existing texts even if introducing human assists to some (reasonable) extent. Four sample texts (totally 40 sentences) selected from existing technical reports and computer menuals. Each of the s ~le texts orresponds to one section or a short chapter in the material. All s ences of each sample texts have been successfully translated into Japanese.</Paragraph>
    <Paragraph position="2"> No pre-editing is done except for three minor modifications to the original text (e.g., &amp;quot;16- or 32- bit&amp;quot; =&gt; &amp;quot;16-bit or 32-bit&amp;quot;). Human assist is limited to resolving ambiguities in the analysis phase. One example is shown in Fig.2.</Paragraph>
  </Section>
class="xml-element"></Paper>