<?xml version="1.0" standalone="yes"?>
<Paper uid="W03-1501">
  <Title>Learning Formulation and Transformation Rules for Multilingual Named Entities</Title>
  <Section position="7" start_page="2" end_page="2" type="concl">
    <SectionTitle>
6 Conclusion and Remarks
</SectionTitle>
    <Paragraph position="0"> This paper proposes corpus-based approaches to extract the formulation rules and the translation/ transliteration rules among multilingual named entities. Simple frequency-based method identifies keywords of named entities for individual languages and their correspondence. The modified tfxidf scheme deals with the issues of abbreviation and compound keyword at a distance.</Paragraph>
    <Paragraph position="1"> Since the corpora are already phrase-aligned, the mined rules cover at least a significant number of instances. That is, they seem to be significant, but further evaluation is needed. Two types of evaluation are being conducted, i.e., direct and indirect approaches. In the former, we will partition the corpora into two parts, one for training and the other one for testing. In the latter, we are integrating our method in a cross language information retrieval system. Given a query consisting of Chinese named entity, the Chinese formulation rules will tell us its type and lexical structures. The transformation rules show which parts should be translated and transliterated. Our previous works on phoneme transliteration is integrated. The transformation result may be submitted to an information retrieval system to access documents in another language. In the ongoing evaluation, the test bed is supported by CLEF (2003). The result will be reported in CLEF2003 after evaluation by CLEF organizer.</Paragraph>
    <Paragraph position="2"> Further applications will be explored in the future and the methodology will be extended to other types of named entities.</Paragraph>
  </Section>
class="xml-element"></Paper>