<?xml version="1.0" standalone="yes"?>
<Paper uid="W00-0505">
  <Title>Towards Translingual Information Access using Portable Information Extraction</Title>
  <Section position="3" start_page="0" end_page="0" type="intro">
    <SectionTitle>
1 Introduction
</SectionTitle>
    <Paragraph position="0"> In this paper, we report on a small study undertaken to demonstrate the feasibility of combining portable information extraction with MT in order to support translingual information access. The goal of our proposed system is to better enable analysts to perform information filtering tasks on foreign language documents.</Paragraph>
    <Paragraph position="1"> This effort was funded by a SBIR Phase I award from the U.S. Army Research Lab, and will be pursued further under the DARPA TIDES initiative.</Paragraph>
    <Paragraph position="2"> Information extraction (IE) systems are designed to extract specific types of information from natural language texts. In order to achieve acceptable accuracy, IE systems need to be tuned for a given topic domain. Since this domain tuning can be labor intensive, recent IE research has focused on developing learning algorithms for training IE system components (cf. Cardie, 1997, for a survey). To date, however, little work has been done on IE systems for languages other than English (though cf. MUC-5, 1994, and MUC-7, 1998, for Japanese IE systems); and, to our knowledge, none of the available techniques for the core task of learning information extraction patterns have been extended or evaluated for multilingual information extraction (though again cf. MUC-7, 1998, where the use of learning techniques for the IE subtasks of named entity recognition and coreference resolution are described).</Paragraph>
    <Paragraph position="3"> Given this situation, the primary objective of our study was to demonstrate the feasibility of using portable--i.e., easily trainable--IE technology on Korean documents, focusing on techniques for learning information extraction patterns. Secondary objectives of the study were to elaborate the analyst scenario and system design.</Paragraph>
  </Section>
class="xml-element"></Paper>