Presentation 2004/10/28
Estimate of Conceptual Vectors for Unregistered Words
Katsuji BESSHO, Masahiro OKU,
PDF Download Page PDF download Page Link
Abstract(in Japanese) (See Japanese page)
Abstract(in English) A conceptual base consisting of conceptual vectors, which are semantic representations of each word generated from co-occurrence patterns with other words, is useful for topic structure extraction from text or information retrieval. However, unregistered words from the conceptual base are not assigned conceptual vectors, and cannot be applied to the processing using conceptual base. We propose a method of estimating the conceptual vectors for unregistered words based on co-occurrence information in the texts. The experimental results of topic segmentation using several articles hi the newspapers show that the proposed method, using estimated vectors, can improve segmentation accuracy.
Keyword(in Japanese) (See Japanese page)
Keyword(in English) Unregistered Word / Conceptual Vector / Topic Segmentation
Paper # NLC2004-20
Date of Issue

Conference Information
Committee NLC
Conference Date 2004/10/28(1days)
Place (in Japanese) (See Japanese page)
Place (in English)
Topics (in Japanese) (See Japanese page)
Topics (in English)
Chair
Vice Chair
Secretary
Assistant

Paper Information
Registration To Natural Language Understanding and Models of Communication (NLC)
Language JPN
Title (in Japanese) (See Japanese page)
Sub Title (in Japanese) (See Japanese page)
Title (in English) Estimate of Conceptual Vectors for Unregistered Words
Sub Title (in English)
Keyword(1) Unregistered Word
Keyword(2) Conceptual Vector
Keyword(3) Topic Segmentation
1st Author's Name Katsuji BESSHO
1st Author's Affiliation NTT Cyber-Solutions Laboratories, NTT Corporation()
2nd Author's Name Masahiro OKU
2nd Author's Affiliation NTT Cyber-Solutions Laboratories, NTT Corporation
Date 2004/10/28
Paper # NLC2004-20
Volume (vol) vol.104
Number (no) 416
Page pp.pp.-
#Pages 6
Date of Issue