<?xml version="1.0" standalone="yes"?>
<Paper uid="P06-2057">
  <Title>Sydney, July 2006. c(c)2006 Association for Computational Linguistics A FrameNet-based Semantic Role Labeler for Swedish</Title>
  <Section position="4" start_page="436" end_page="436" type="relat">
    <SectionTitle>
1.2 Related Work
</SectionTitle>
    <Paragraph position="0"> Since training data is often a scarce resource for most languages other than English, a wide range of methods have been proposed to reduce the need for manual annotation. Many of these have relied on existing resources for English and a transfer method based on word alignment in a parallel corpus to automatically create an annotated corpus in a new language. Although these data are typically quite noisy, they have been used to train automatic systems.</Paragraph>
    <Paragraph position="1"> For the particular case of transfer of FrameNet annotation, there have been a few projects that have studied transfer methods and evaluated the quality of the automatically produced corpus. Johansson and Nugues (2005) applied the word-based methods of Yarowsky et al. (2001) and obtained promising results. Another recent effort (Pado and Lapata, 2005) demonstrates that deeper linguistic information, such as parse trees in the source and target language, is very beneficial for the process of FrameNet annotation transfer.</Paragraph>
    <Paragraph position="2"> A rather different method to construct bilingual semantic role annotation is the approach taken by BiFrameNet (Fung and Chen, 2004). In that work, annotated structures in a new language (in that case Chinese) are produced by mining for similar structures rather than projecting them via parallel corpora.</Paragraph>
  </Section>
class="xml-element"></Paper>