Presentation 2017-06-10
A Study on Acquisition of Associative Word for Topic Transition in a Chat Dialogue System
Kazuki Asami, Toru Sugimoto,
PDF Download Page PDF download Page Link
Abstract(in Japanese) (See Japanese page)
Abstract(in English) Chat dialogue systems need to have a topic transition function in order not to let users get tired. In this paper, we focus on association as a method to decide topics after topic transition, and study acquisition of associative words that realize natural topic transition. In order to clarify properties and conditions of association used for topic transition, we collect associative words by free association with topic words as stimulus words. We also collect related words that have large cosine similarity with topic words based on word vectors acquired using word2vec. Next, we classify the collected associative words and related words according to the semantic relation with topic words, and compare them. Then we conduct evaluation experiments to investigate whether these associative words and related words are suitable for topic transition in dialogue, and consider a method to automatically obtain associative words that realize natural topic transition.
Keyword(in Japanese) (See Japanese page)
Keyword(in English) Association / Chat Dialogue System / Topic Transition / word2vec
Paper # TL2017-11,NLC2017-11
Date of Issue 2017-06-02 (TL, NLC)

Conference Information
Committee NLC / TL
Conference Date 2017/6/9(2days)
Place (in Japanese) (See Japanese page)
Place (in English) Tottori University
Topics (in Japanese) (See Japanese page)
Topics (in English) Application of natural language proessing and linguistic analysis, and general topic of NLP
Chair Hiroshi Kanayama(IBM) / Masami Suzuki(KDDI Research)
Vice Chair Takeshi Sakaki(Hottolink) / Kazutaka Shimada(Kyushu Inst. of Tech.) / Tadahisa Kondo(Kogakuin Univ.) / Chiaki Kubomura(Yamano College of Aesthetics)
Secretary Takeshi Sakaki(Ryukoku Univ.) / Kazutaka Shimada(NTT) / Tadahisa Kondo(Kobe Gakuin Univ.) / Chiaki Kubomura(Kyoto Univ.)
Assistant Mitsuo Yoshida(Toyohashi Univ. of Tech.) / Takeshi Kobayakawa(NICT) / Nobuyuki Jincho(Waseda Univ.) / Noriaki Takada(Ferris Univ.)

Paper Information
Registration To Technical Committee on Natural Language Understanding and Models of Communication / Technical Committee on Thought and Language
Language JPN
Title (in Japanese) (See Japanese page)
Sub Title (in Japanese) (See Japanese page)
Title (in English) A Study on Acquisition of Associative Word for Topic Transition in a Chat Dialogue System
Sub Title (in English)
Keyword(1) Association
Keyword(2) Chat Dialogue System
Keyword(3) Topic Transition
Keyword(4) word2vec
1st Author's Name Kazuki Asami
1st Author's Affiliation Shibaura Institute of Technology(Shibaura Inst. Tech.)
2nd Author's Name Toru Sugimoto
2nd Author's Affiliation Shibaura Institute of Technology(Shibaura Inst. Tech.)
Date 2017-06-10
Paper # TL2017-11,NLC2017-11
Volume (vol) vol.117
Number (no) TL-81,NLC-82
Page pp.pp.59-64(TL), pp.59-64(NLC),
#Pages 6
Date of Issue 2017-06-02 (TL, NLC)