<?xml version="1.0" standalone="yes"?>
<Paper uid="I05-2027">
  <Title>Machine Learning Approach To Augmenting News Headline Generation</Title>
  <Section position="9" start_page="158" end_page="158" type="concl">
    <SectionTitle>
6 Conclusions and Future work
</SectionTitle>
    <Paragraph position="0"> The results of our experiment have shown the TFTrim system (the simplest of the three Topiary-style headline generators examined in this paper) is the most appropriate headline approach because it yields high quality short summaries and, unlike the Topiary and HybridTrim systems, it requires no prior training. This is an interesting result as it shows that a simple tf weighting scheme can produce as good, if not better, topic descriptors than the statistical UTD method employed by the University of Maryland and our own statistical/linguistic approach to topic label identification.</Paragraph>
    <Paragraph position="1"> In future work, we intend to proceed by improving the sentence compression procedure described in this paper. We are currently working on the use of term frequency information as a means of improving the performance of the Hedge Trimmer algorithm by limiting the elimination of important parse tree components during sentence compression.</Paragraph>
  </Section>
class="xml-element"></Paper>