<?xml version="1.0" standalone="yes"?>
<Paper uid="W05-1304">
  <Title>Proceedings of the ACL-ISMB Workshop on Linking Biological Literature, Ontologies and Databases: Mining Biological Semantics, pages 25-31, Detroit, June 2005. c(c)2005 Association for Computational Linguistics A Machine Learning Approach to Acronym Generation</Title>
  <Section position="3" start_page="25" end_page="25" type="intro">
    <SectionTitle>
2 Acronym Generation as a Sequence
</SectionTitle>
    <Paragraph position="0"/>
    <Section position="1" start_page="25" end_page="25" type="sub_section">
      <SectionTitle>
Labeling Problem
</SectionTitle>
      <Paragraph position="0"> Given the definition (expanded form), the mechanism of acronym generation can be regarded as the task of selecting the appropriate action on each letter in the definition.</Paragraph>
      <Paragraph position="1"> Figure 1 illustrates an example, where the definition is &amp;quot;Duck interferon gamma&amp;quot; and the generated acronym is &amp;quot;DuIFN-gamma&amp;quot;. The generation proceeds as follows: The acronym generator outputs the first two letters unchanged and skips the following three letters. Then the generator capitalizes 'i' and skip the following four letters...</Paragraph>
      <Paragraph position="2"> By assuming that an acronym is made up of alpha-numeric letters, spaces and hyphens, the actions being taken by the generator are classified into the following five classes.</Paragraph>
    </Section>
  </Section>
class="xml-element"></Paper>