<?xml version="1.0" standalone="yes"?>
<Paper uid="A97-1013">
  <Title>Developing a hybrid NP parser</Title>
  <Section position="7" start_page="85" end_page="85" type="evalu">
    <SectionTitle>
7 Experiments and results
</SectionTitle>
    <Paragraph position="0"> We tested linguistic, statistical and hybrid language models, using the CG-2 parser (Tapanainen, 1996) and the relaxation labelling algorithm described in Section 2.</Paragraph>
    <Paragraph position="1"> The statistical models were obtained from a training corpus of 218,000 words of journalese, syntactically annotated using the linguistic parser (see above).</Paragraph>
    <Paragraph position="2"> Although the linguistic CG-2 parser does not disambiguate completely, it seems to have an almost perfect recall (cf. Table 1 below), and the noise introduced by the remaining ambiguity is assumed to be sufficiently lower than the signal, following the idea used in (Yarowsky, 1992).</Paragraph>
    <Paragraph position="3"> The collected statistics were bigram and trigram occurrences.</Paragraph>
    <Paragraph position="4"> The algorithms and models were tested against a hand-disambiguated benchmark corpus of over 6,500 words.</Paragraph>
    <Paragraph position="5"> We measure the performance of the different models in terms of recall and precision. Recall is the percentage of words that get the correct tag among the tags proposed by the system. Precision is the percentage of tags proposed by the system that are correct.</Paragraph>
  </Section>
class="xml-element"></Paper>