A Novel Japanese Anaphora Resolution Method Using Deep Cases

Takumi Kawasaki, Masaomi Kimura

研究成果: Conference contribution

抜粋

Anaphora resolution resolves a reference of a pronoun to a noun in a phrase. It is an active area of research, especially in text mining, because any result can be missed due to a mean of sentences including abbreviations or pronouns. An existing system, called zero anaphora analysis, cannot guarantee accurate results of detecting and complementing abbreviated pronouns as for Japanese sentence. Previous studies of anaphora resolution used surface case, which represents the semantic relationships between verbs and nouns. However, they analyzed only at the surface information of the emerging particle; therefore, it is difficult to obtain an appropriate result. Obviously, not only the analysis of superficial sentences but also the deep case should be used to address the problem. In this paper, we propose a new anaphora resolution method by using deep case.

元の言語English
ホスト出版物のタイトルProceedings - 2017 International Symposium on Computer Science and Intelligent Controls, ISCSIC 2017
出版者Institute of Electrical and Electronics Engineers Inc.
ページ129-134
ページ数6
2018-February
ISBN(電子版)9781538629413
DOI
出版物ステータスPublished - 2018 2 16
イベント1st International Symposium on Computer Science and Intelligent Controls, ISCSIC 2017 - Budapest, Hungary
継続期間: 2017 10 202017 10 22

Other

Other1st International Symposium on Computer Science and Intelligent Controls, ISCSIC 2017
Hungary
Budapest
期間17/10/2017/10/22

ASJC Scopus subject areas

  • Computer Science Applications
  • Artificial Intelligence

フィンガープリント A Novel Japanese Anaphora Resolution Method Using Deep Cases' の研究トピックを掘り下げます。これらはともに一意のフィンガープリントを構成します。

  • これを引用

    Kawasaki, T., & Kimura, M. (2018). A Novel Japanese Anaphora Resolution Method Using Deep Cases. : Proceedings - 2017 International Symposium on Computer Science and Intelligent Controls, ISCSIC 2017 (巻 2018-February, pp. 129-134). [8294173] Institute of Electrical and Electronics Engineers Inc.. https://doi.org/10.1109/ISCSIC.2017.10