<?xml version="1.0" standalone="yes"?>
<Paper uid="C92-2067">
  <Title>A New Quantitative Quality Measure for Machine 'lYanslation Systems</Title>
  <Section position="2" start_page="0" end_page="0" type="intro">
    <SectionTitle>
1. Introduction
</SectionTitle>
    <Paragraph position="0"> There are several reasons performance measure is required while building machine translation systems. (1).</Paragraph>
    <Paragraph position="1"> Potential customers need to be able to compare the performance of different systems. (2). System designers would like to keep abreast of the current system performauce, and make sure the system keeps on improving, not subject to tbeflip-flop problem. (The flip-flop problem is caused when system designers try to fix soore problem of the system without a thorough test. While that problem is solved, other problems may pop up. This kind of problem becomes more serious when the system scales up.) (3). The measure feedback will highlight correct research direction. (4). In a parameterized system,</Paragraph>
  </Section>
class="xml-element"></Paper>