<?xml version="1.0" standalone="yes"?>
<Paper uid="C02-1019">
  <Title>The Grammatical Function Analysis between Korean Adnoun Clause and Noun Phrase by Using Support Vector Machines</Title>
  <Section position="8" start_page="2" end_page="2" type="concl">
    <SectionTitle>
5 Conclusion and Future works
</SectionTitle>
    <Paragraph position="0"> Adnoun clause is a typical complex sentence structure of Korean. There are various types of grammatical relations between an adnoun clause and its relevant noun phrase. Unlike in between general content words and modifying clauses where their grammatical relations can be easily extrated in terms of various grammatical characteristics by the functional morphemes, the functional morphemes are omitted in a noun phrase when it is modified by an adnoun clause.</Paragraph>
    <Paragraph position="1"> This omission makes it difficult to characterize their grammatical relation.</Paragraph>
    <Paragraph position="2"> In this paper, we used SVM to take care of this problem and analyze the relation between noun phrase and adnoun clause. We reflected context information by using the previous word of the verb in adnoun clauses as feature. Context information helped the grammatical function analysis between adnoun clause and the head noun. The SVM can also handle the sparse data problem as the backed-off model does. We acquired overall accuracy of 90.8%, which is obviously an improvement from the previous works.</Paragraph>
    <Paragraph position="3"> In the future, we plan to compare with other machine learning methods and to enhance our system by using a publicly available Korean thesaurus to increases general accuracy. More data needs to be collected for further performance improvement. We will also work on utilizing the proposed model in some partial parsing problem.</Paragraph>
  </Section>
class="xml-element"></Paper>