<?xml version="1.0" standalone="yes"?>
<Paper uid="W06-2922">
  <Title>Experiments with a Multilanguage Non-Projective Dependency Parser</Title>
  <Section position="4" start_page="166" end_page="166" type="intro">
    <SectionTitle>
2 Features
</SectionTitle>
    <Paragraph position="0"> Columns from input data were used as follows.</Paragraph>
    <Paragraph position="1"> LEMMA was used in features whenever available, otherwise the FORM was used. For Danish, German and Swedish the Snowball stemmer (Porter 2001) was used to generate a value for LEMMA. This use of stemming slightly improved both accuracy and performance.</Paragraph>
    <Paragraph position="2"> Only CPOSTAG were used. PHEAD/PDEPREL were not used.</Paragraph>
    <Paragraph position="3"> FEATS were used to extract a single token combining gender, number, person and case, through a language specific algorithm.</Paragraph>
    <Paragraph position="4"> The selection of features to be used in the parser is controlled by a number of parameters. For example, the parameter PosFeatures determines for which tokens the POS tag will be included in the context, PosLeftChildren determines how many left outermost children of a token to consider, PastActions tells how many previous actions to include as features.</Paragraph>
    <Paragraph position="5"> The settings used in the submitted runs are listed below and configure the parser for not using any word forms. Positive numbers refer to input tokens, negative ones to token on the stack.</Paragraph>
    <Paragraph position="6">  The context for POS tags consisted of 1 token left and 3 tokens to the right of the focus words, except for Czech and Chinese were 2 tokens to the left and 4 tokens to the right were used. These values were chosen by performing experiments on the training data, using 10% of the sentences as held-out data for development.</Paragraph>
  </Section>
class="xml-element"></Paper>