Building a Paraphrase Dataset for Language Learners : Paraphrase Extraction with Document Structure Model

https://doi.org/10.24546/81006699

このアイテムのアクセス数:0件（2024-04-18 13:34 集計）

ファイル	フォーマット	サイズ	閲覧回数	説明
81006699 (fulltext)	pdf	5.98 MB	4

メタデータID	81006699
アクセス権	open access
出版タイプ	Version of Record
タイトル	Building a Paraphrase Dataset for Language Learners : Paraphrase Extraction with Document Structure Model
著者	著者名 Wang, Wangjie
収録物名	Learner Corpus Studies in Asia and the World
巻(号)	2
ページ	177-184
出版者	神戸大学国際コミュニケーションセンター
刊行日	2014-05-31
公開日	2014-06-25
抄録（自由利用可）	Paraphrasing techniques, especially data-driven approaches require plenty of language resources of paraphrases to build the models. Most of traditional methods for paraphrase extraction only consider the semantic similarity of target sentences. However, in the comparable corpus, especially those consisting of parallel documents, there are some common patterns in the context of sentences, which can be used as effective identifiers. We studied the problem of paraphrase extraction from parallel documents, and proposed a Document Structure (DS) Model based scheme. This model was applied to the International Corpus Network of Asian Learners of English (ICNALE). We evaluated the performance of the proposed method, and collected a paraphrase aligned dataset from ICNALE with this method
カテゴリ	Learner Corpus Studies in Asia and the World ＞ 2号（2014-05-31）
カテゴリ	紀要論文

資源タイプ	departmental bulletin paper
言語	English (英語)
ISSN	2187-6746　OPACで所蔵を検索　 CiNiiで学外所蔵を検索
関連情報	URI http://www.solac.kobe-u.ac.jp/