<?xml version="1.0" standalone="yes"?>
<Paper uid="P98-2159">
  <Title>An Efficient Parallel Substrate for Typed Feature Structures on Shared Memory Parallel Machines</Title>
  <Section position="6" start_page="972" end_page="973" type="evalu">
    <SectionTitle>
2. Results and comparison with other sequen-
</SectionTitle>
    <Paragraph position="0"> tial parsing systems are given in Table 3. Its speedup is shown in Figure 7. From the figure, we observe that the maximum speedup reaches up to 10.9 times and its parsing time is 1776 msec per sentence.</Paragraph>
    <Section position="1" start_page="972" end_page="973" type="sub_section">
      <SectionTitle>
4.3 Discussion
</SectionTitle>
      <Paragraph position="0"> In both parsers, parsing time reaches a level required by real-time applications, though we used computationally expensive grammar formalisms, i.e. HPSG with reasonable coverage and accuracy. This shows the feasibility of our 7This sample grammar is converted to LiLFeS style half automatically.</Paragraph>
      <Paragraph position="1">  framework for the goal to provide a parallel programming environment for real-time NLP. In addition, our parallel HPSG parsers are considerably more efficient than other sequential HPSG parsers.</Paragraph>
      <Paragraph position="2"> However, the speed-up is not proportional to the number of processors. We think that this is because the parallelism extracted in our parsing algorithm is not enough. Figure 8 shows the log of parsing Japanese sentences by the CKY-style parser. The black lines indicate when a processor is busy. One can see that many processors are frequently idle.</Paragraph>
      <Paragraph position="3"> We think that this idle time does not suggest that parallel NLP systems are useless. On the contrary, this suggest that parallel NLP systems have many possibilities. If we introduce semantic processing for instance, overall processing time may not change because the idle time is used for semantic processing. Another possibility is the use of parallel NLP systems as a server. Even if we feed several sentences at a time, throughput will not change, because the idle time is used for parsing different sentences.</Paragraph>
    </Section>
  </Section>
class="xml-element"></Paper>