<?xml version="1.0" standalone="yes"?>
<Paper uid="W04-3203">
  <Title>Induction of Greedy Controllers for Deterministic Treebank Parsers</Title>
  <Section position="8" start_page="0" end_page="0" type="concl">
    <SectionTitle>
7 Conclusions
</SectionTitle>
    <Paragraph position="0"> Deterministic unlexicalized statistical parsers have surprisingly good accuracy and coverage, considering their speed and simplicity. The best parsers reported here have almost complete coverage, outperform basic PCFGs, and are roughly as fast as taggers. We described an approach to statistical parsing based on induction of stochastic automata. We de ned Markov parsing models, described how to estimate parameters for them, and showed how the deterministic parsers we implemented are greedy versions of MPM parsers. We found that for greedy parsing, bottom-up parsing strategies seem to have a small advantage over top-down.</Paragraph>
  </Section>
class="xml-element"></Paper>