<?xml version="1.0" standalone="yes"?>
<Paper uid="A00-1033">
  <Title>A Divide-and-Conquer Strategy for Shallow Parsing of German Free Texts</Title>
  <Section position="10" start_page="245" end_page="245" type="concl">
    <SectionTitle>
6 Conclusion and future work
</SectionTitle>
    <Paragraph position="0"> We have presented a divide-and-conquer strategy for shallow analysis of German texts which is supported by means of powerful morphological processing, efficient POS-filtering and named entity recognition. Especially for the divide-and-conquer parsing strategy we obtained an F-measure of 87.14% on unseen data. Our shallow parsing strategy has a high degree of modularity which allows the integration of the domain-independent sentence recognition part with arbitrary domain-dependent sub-components (e.g., specific named entity finders and fragment recognizers).</Paragraph>
    <Paragraph position="1"> Considered from an application-oriented point of view, our main experience is that even if we are only interested in some parts of a text (e.g., only in those linguistic entities which verbalize certain aspects of a domain-concept) we have to unfold the structural relationship between all elements of a large enough area (a paragraph or more) up to a certain level of depth in which the relevant information is embedded. Beside continuing the improvement of the whole approach we also started investigations towards the integration of deep processing into the DC-PARSER. The core idea is to call a deep parser only to the separated field elements which contain sequences of simple NPs and PPs (already determined by the shallow parser). Thus seen the shallow parser is used as an efficient preprocessor for dividing a sentence into syntactically valid smaller units, where the deep parser's task would be to identify the exact constituent structure only on demand.</Paragraph>
  </Section>
class="xml-element"></Paper>