<?xml version="1.0" standalone="yes"?>
<Paper uid="M93-1013">
  <Title>SAID FRIDAY IT WILL SET U P A JOINT VENTUR E IN TAIWA N WITH A LOCAL CONCER N AND A JAPANESE TRADING HOUS E TO PRODUCE GOLF CLUBS</Title>
  <Section position="7" start_page="144" end_page="144" type="concl">
    <SectionTitle>
FUTURE WORK
</SectionTitle>
    <Paragraph position="0"> Alembic is still on the early end of its maturation curve . Our grammar, for example, was undergoing constan t development up until the fmal runs--unlike many other MUC sites, we did not enter this endeavor with anythin g remotely close to a general grammar of English, or even of the so-called noun and verb groups . Throughout thi s time period, basic system development was thus responsible for stealing precious time away from domai n engineering.</Paragraph>
    <Paragraph position="1"> One benefit of all our MUC-5 development, is that Alembic has reached a reasonably stable state. We now have a flexible framework within which to explore data extraction tasks such as EJV in MUC-5 . We are especially encouraged by our recall/precision profile over the last few weeks of development . Our daily tests revealed a consistently high precision level that remained stable even with daily improvements in recall.</Paragraph>
    <Paragraph position="2"> We are also pleased by our preliminary experiences with the new inference component . Among MUC-5 participants, Alembic is distinguished by its ability to accommodate strategies that place a high premium on full parsing (the lexical inference rules) alongside strategies more geared towards fragment parsing (the event mergin g rules). We look forward to exploiting this new flexibility as we continue informal self-evaluation on the MUC- 5 task, and as we explore future data extraction problems.</Paragraph>
    <Paragraph position="3"> 14 5</Paragraph>
  </Section>
class="xml-element"></Paper>