A knowledge processing neural network based on automatic concept hierarchization

Masahiro Saito, Masafumi Hagiwara

研究成果: Conference contribution

抄録

In this paper, we propose a knowledge processing neural network which is capable of inductive and deductive inference. The proposed network looks up relations between words in a concept dictionary and co-occurrence dictionary. First, the proposed network divides sentences into the subject words and the other words. Then these words are input into two-layer network. Second, hierarchical structure is composed using concept dictionary. Third, the network induces general knowledge from individual knowledge. We added a function to respond to questions in natural language with "Yes/No/I don't know" in order to confirm the validity of proposed network by evaluating the quantity of correct answers.

本文言語English
ホスト出版物のタイトルNeural Information Processing - 14th International Conference, ICONIP 2007, Revised Selected Papers
ページ539-548
ページ数10
PART 2
DOI
出版ステータスPublished - 2008 10 24
イベント14th International Conference on Neural Information Processing, ICONIP 2007 - Kitakyushu, Japan
継続期間: 2007 11 132007 11 16

出版物シリーズ

名前Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
番号PART 2
4985 LNCS
ISSN(印刷版)0302-9743
ISSN(電子版)1611-3349

Other

Other14th International Conference on Neural Information Processing, ICONIP 2007
CountryJapan
CityKitakyushu
Period07/11/1307/11/16

ASJC Scopus subject areas

  • Theoretical Computer Science
  • Computer Science(all)

フィンガープリント 「A knowledge processing neural network based on automatic concept hierarchization」の研究トピックを掘り下げます。これらがまとまってユニークなフィンガープリントを構成します。

引用スタイル