Presentation 2008-02-07
Use of Syntactic Pieces to Extract Opinions Without Identifying Domains
Suguru AOKI, Kazuhide YAMAMOTO,
PDF Download Page PDF download Page Link
Abstract(in Japanese) (See Japanese page)
Abstract(in English) In this paper, we propose a opinion extraction method using a notion of syntactic piece. Syntactic piece is a minimum unit of structure. It is easy to use like n-gram and it has a part of syntactic structure. In the task of opinion extraction, syntactic piece can be high accuracy classification. Because, it contain both object of evaluation and sentiment expression. Therefore, we do not need to make dictionary for each domain. First, we made syntactic piece dictionary tagged positive or negative using training data, and classified sentence. Next, we generalized and extended dictionary to improve recall. As the result, the system performed 80% precision and 50% recall for multiple domains.
Keyword(in Japanese) (See Japanese page)
Keyword(in English) syntactic piece / opinion extraction / domain dependency / sentence classification
Paper # NLC2007-88
Date of Issue

Conference Information
Committee NLC
Conference Date 2008/1/31(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) Use of Syntactic Pieces to Extract Opinions Without Identifying Domains
Sub Title (in English)
Keyword(1) syntactic piece
Keyword(2) opinion extraction
Keyword(3) domain dependency
Keyword(4) sentence classification
1st Author's Name Suguru AOKI
1st Author's Affiliation Nagaoka University of Technology()
2nd Author's Name Kazuhide YAMAMOTO
2nd Author's Affiliation Nagaoka University of Technology
Date 2008-02-07
Paper # NLC2007-88
Volume (vol) vol.107
Number (no) 480
Page pp.pp.-
#Pages 6
Date of Issue