Presentation 2007/7/17
Attribute-value extraction from description of exhibits for facetted search in net auction system
Jun NISHIMURA, Rintaro MIYAZAKI, Naoto MAEDA, Tatsunori MORI, Shore O, Yusuke ISHIKAWA, Hiroyuki KOBAYASHI, Yuya TANAKA, Fuyuko KIDO,
PDF Download Page PDF download Page Link
Abstract(in Japanese) (See Japanese page)
Abstract(in English) In order to achieve flexible facetted search for descriptions of exhibits in net auction system, in this paper, we studied automated extraction of attributes and their values, which appear in those descriptions, based on a machine learning technique. First of all, we examined a set of attributes that should be indexed for the facetted search. Especially, we focused on attributes that can be annotated stably by different annotators, and that are needed for search We also studied a way to deal with the diversity of attributes and values in descriptions of exhibits. When surface expressions are directly used as one of features, the result of learning may be unwillingly over-fitted to training corpora, and consequently the performance of information extraction will be degraded. Therefore, we introduced the category information of a thesaurus, which does not depend on surface expression directly, and examined the effectiveness of the feature. With regard to the extraction method, we adopted a standard character-based chunking method, which are usually used for named entity extraction.
Keyword(in Japanese) (See Japanese page)
Keyword(in English) net auction / attribute / information extraction / chunking
Paper # NLC2007-27
Date of Issue

Conference Information
Committee NLC
Conference Date 2007/7/17(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) Attribute-value extraction from description of exhibits for facetted search in net auction system
Sub Title (in English)
Keyword(1) net auction
Keyword(2) attribute
Keyword(3) information extraction
Keyword(4) chunking
1st Author's Name Jun NISHIMURA
1st Author's Affiliation Yokohama National University()
2nd Author's Name Rintaro MIYAZAKI
2nd Author's Affiliation Yokohama National University
3rd Author's Name Naoto MAEDA
3rd Author's Affiliation Yokohama National University
4th Author's Name Tatsunori MORI
4th Author's Affiliation Yokohama National University
5th Author's Name Shore O
5th Author's Affiliation Yahoo Japan Corporation
6th Author's Name Yusuke ISHIKAWA
6th Author's Affiliation Yahoo Japan Corporation
7th Author's Name Hiroyuki KOBAYASHI
7th Author's Affiliation Yahoo Japan Corporation
8th Author's Name Yuya TANAKA
8th Author's Affiliation Yahoo Japan Corporation
9th Author's Name Fuyuko KIDO
9th Author's Affiliation Yahoo Japan Corporation
Date 2007/7/17
Paper # NLC2007-27
Volume (vol) vol.107
Number (no) 158
Page pp.pp.-
#Pages 6
Date of Issue