Topic Dependent Language Model based on On-Line Voting

• Item type: SIG Technical Reports(1)
• 公開日: 2009-12-14
• タイトル: Topic Dependent Language Model based on On-Line Voting
• タイトル / 言語: en
• タイトル: Topic Dependent Language Model based on On-Line Voting
• 言語: eng
• キーワード / 主題Scheme: Other
• キーワード / 主題: 【Session-3 言語モデル】
• 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: Department of Information and Computer Sciences, Toyohashi University of Technology
• 著者所属: Information and Media Center, Toyohashi University of Technology
• 著者所属: Department of Information and Computer Sciences, Toyohashi University of Technology
• 著者所属(英): en
• 著者所属(英): Department of Information and Computer Sciences, Toyohashi University of Technology
• 著者所属(英): en
• 著者所属(英): Information and Media Center, Toyohashi University of Technology
• 著者所属(英): en
• 著者所属(英): Department of Information and Computer Sciences, Toyohashi University of Technology
• 著者名: Welly, Naptali; Masatoshi, Tsuchiya; Seiichi, Nakagawa
• 著者名(英): Welly, Naptali; Masatoshi, Tsuchiya; Seiichi, Nakagawa
• 論文抄録 / 内容記述タイプ: Other
• 論文抄録 / 内容記述: In this paper, we propose an alternative approach to a topic dependent language model (LM), where the topic is decided by voting in an unsupervised manner. Latent Semantic Analysis (LSA) is employed to reveal hidden (latent) relations among nouns in the context word sequence. To decide the topic of an event, a fixed size word history sequence (window) is observed, and voting is then carried out based on noun class occurrences weighted by a confidence measure. Experiments on the Wall Street Journal corpus and Mainichi Shimbun (Japanese newspaper) corpus show that our proposed method gives better perplexity than the comparative baselines, including a word-based/class-based n-gram LM, their interpolated LM, a cache-based LM, and the Latent Dirichlet Allocation (LDA)-based topic dependent LM.
• 論文抄録(英) / 内容記述タイプ: Other
• 論文抄録(英) / 内容記述: In this paper, we propose an alternative approach to a topic dependent language model (LM), where the topic is decided by voting in an unsupervised manner. Latent Semantic Analysis (LSA) is employed to reveal hidden (latent) relations among nouns in the context word sequence. To decide the topic of an event, a fixed size word history sequence (window) is observed, and voting is then carried out based on noun class occurrences weighted by a confidence measure. Experiments on the Wall Street Journal corpus and Mainichi Shimbun (Japanese newspaper) corpus show that our proposed method gives better perplexity than the comparative baselines, including a word-based/class-based n-gram LM, their interpolated LM, a cache-based LM, and the Latent Dirichlet Allocation (LDA)-based topic dependent LM.
• 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: AN10442647
• 書誌情報: 音声言語情報処理（SLP）; 巻 2009-SLP-79, 号 8, p. 1-6, 発行日 2009-12-14
• 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