<?xml version="1.0" standalone="yes"?>
<Paper uid="M98-1014">
  <Title>FACILE: DESCRIPTION OF THE NE SYSTEM USED FOR MUC-7</Title>
  <Section position="3" start_page="0" end_page="0" type="intro">
    <SectionTitle>
INTRODUCTION
</SectionTitle>
    <Paragraph position="0"> In this paper, we describe the system used by the UMIST team as members of the FACILE consortium, to undertake the NE task in MUC-7. The main characteristics of this system employed are as follows: #0F it is rule-based #0F its rule formalism supports context-sensitive partial parsing #0F rules may use pattern-matching-style iteration operators #0F the notation is much more readable than classic pattern-matching languages #0F rules can be assigned an explicit weight which is used in choosing between competing analyses. #0F there is a method for identifying name-strings as coreferential with longer variants in the same text #0F the system does not employ learning techniques The development of the system began only about 20 months ago, so it has not been used in any previous comparable trials. We looked forward to slightly higher scores than we obtained in the formal run, because at the dry run stage, we had obtained almost identical scores to our best results with training data.</Paragraph>
    <Paragraph position="1"> In the rest of this paper, we first give some background on the context in which the system used in the MUC-7 NE task was developed. We then outline its internal structure, concentrating on the rule notation which is its most salient feature. An evaluation of its performance in the task then follows, before concluding with some speculation on the extent to which the approach adopted is susceptible to further improvement.</Paragraph>
  </Section>
class="xml-element"></Paper>