Dual Decomposition for Chinese Semantic Role Labeling

• Item type: SIG Technical Reports(1)
• 公開日: 2012-01-13
• タイトル: Dual Decomposition for Chinese Semantic Role Labeling
• タイトル / 言語: en
• タイトル: Dual Decomposition for Chinese Semantic Role Labeling
• 言語: eng
• キーワード / 主題Scheme: Other
• キーワード / 主題: 辞書・構文・意味解析
• 資源タイプ / 資源タイプ識別子: http://purl.org/coar/resource_type/c_18gh
• 資源タイプ: technical report
• 著者所属: Nara Institute of Science and Technology
• 著者所属: Nara Institute of Science and Technology
• 著者所属: Nara Institute of Science and Technology
• 著者所属(英): en
• 著者所属(英): Nara Institute of Science and Technology
• 著者所属(英): en
• 著者所属(英): Nara Institute of Science and Technology
• 著者所属(英): en
• 著者所属(英): Nara Institute of Science and Technology
• 著者名: Yanyan, Luo; Masayuki, Asahara; Yuji, Matsumoto
• 著者名(英): Yanyan, Luo; Masayuki, Asahara; Yuji, Matsumoto
• 論文抄録 / 内容記述タイプ: Other
• 論文抄録 / 内容記述: Semantic role Labeling(SRL) is gaining more attention as it benefits a wide range of natural language processing applications. In the past few years, dependency-based SRL has attracted much attention. However, the bottleneck for Chinese dependency parsing severely limits the ability to solve Chinese SRL. Chinese shallow parsing has better results, but only using this kind of features cannot get ideal results for SRL over word units. This paper describes a dual decomposition method to get the final inference, which can leverage both the dependency parsing information and shallow parsing information. In the experiments, the proposed model achieved much performance improvements.
• 論文抄録(英) / 内容記述タイプ: Other
• 論文抄録(英) / 内容記述: Semantic role Labeling(SRL) is gaining more attention as it benefits a wide range of natural language processing applications. In the past few years, dependency-based SRL has attracted much attention. However, the bottleneck for Chinese dependency parsing severely limits the ability to solve Chinese SRL. Chinese shallow parsing has better results, but only using this kind of features cannot get ideal results for SRL over word units. This paper describes a dual decomposition method to get the final inference, which can leverage both the dependency parsing information and shallow parsing information. In the experiments, the proposed model achieved much performance improvements.
• 書誌レコードID / 収録物識別子タイプ: NCID
• 書誌レコードID / 収録物識別子: AN10114171
• 書誌情報: 研究報告情報基礎とアクセス技術（IFAT）; 巻 2012-IFAT-105, 号 5, p. 1-7, 発行日 2012-01-13
• Notice: SIG Technical Reports are nonrefereed and hence may later appear in any journals, conferences, symposia, etc.
• 出版者 / 言語: ja
• 出版者: 情報処理学会