{"updated":"2025-01-22T08:45:26.291032+00:00","metadata":{"_oai":{"id":"oai:ipsj.ixsq.nii.ac.jp:00047941","sets":["1164:4179:4194:4199"]},"path":["4199"],"owner":"1","recid":"47941","title":["無限混合ディリクレ文書モデル"],"pubdate":{"attribute_name":"公開日","attribute_value":"2006-03-27"},"_buckets":{"deposit":"7cb2a7db-9b6c-48a8-a372-05f2d3c5c3a8"},"_deposit":{"id":"47941","pid":{"type":"depid","value":"47941","revision_id":0},"owners":[1],"status":"published","created_by":1},"item_title":"無限混合ディリクレ文書モデル","author_link":["0","0"],"item_titles":{"attribute_name":"タイトル","attribute_value_mlt":[{"subitem_title":"無限混合ディリクレ文書モデル"},{"subitem_title":"Infinite Dirichlet Mixtures in Text Modeling","subitem_title_language":"en"}]},"item_type_id":"4","publish_date":"2006-03-27","item_4_text_3":{"attribute_name":"著者所属","attribute_value_mlt":[{"subitem_text_value":"ATR 音声言語コミュニケーション研究所 音声言語処理研究室"},{"subitem_text_value":"ATR 音声言語コミュニケーション研究所 音声言語処理研究室"}]},"item_4_text_4":{"attribute_name":"著者所属(英)","attribute_value_mlt":[{"subitem_text_value":"ATR Spoken Language Communication Research Laboratories","subitem_text_language":"en"},{"subitem_text_value":"ATR Spoken Language Communication Research Laboratories","subitem_text_language":"en"}]},"item_language":{"attribute_name":"言語","attribute_value_mlt":[{"subitem_language":"jpn"}]},"item_publisher":{"attribute_name":"出版者","attribute_value_mlt":[{"subitem_publisher":"情報処理学会","subitem_publisher_language":"ja"}]},"publish_status":"0","weko_shared_id":-1,"item_file_price":{"attribute_name":"Billing file","attribute_type":"file","attribute_value_mlt":[{"url":{"url":"https://ipsj.ixsq.nii.ac.jp/record/47941/files/IPSJ-NL06172007.pdf"},"date":[{"dateType":"Available","dateValue":"2008-03-27"}],"format":"application/pdf","billing":["billing_file"],"filename":"IPSJ-NL06172007.pdf","filesize":[{"value":"411.9 kB"}],"mimetype":"application/pdf","priceinfo":[{"tax":["include_tax"],"price":"660","billingrole":"5"},{"tax":["include_tax"],"price":"330","billingrole":"6"},{"tax":["include_tax"],"price":"0","billingrole":"23"},{"tax":["include_tax"],"price":"0","billingrole":"44"}],"accessrole":"open_date","version_id":"2a34f9f5-7fdc-421b-a6be-4c139e7ac6a7","displaytype":"detail","licensetype":"license_note","license_note":"Copyright (c) 2006 by the Information Processing Society of Japan"}]},"item_4_creator_5":{"attribute_name":"著者名","attribute_type":"creator","attribute_value_mlt":[{"creatorNames":[{"creatorName":"持橋, 大地"},{"creatorName":"菊井, 玄一郎"}],"nameIdentifiers":[{}]}]},"item_4_creator_6":{"attribute_name":"著者名(英)","attribute_type":"creator","attribute_value_mlt":[{"creatorNames":[{"creatorName":"Daichi, Mochihashi","creatorNameLang":"en"},{"creatorName":"Genichiro, Kikui","creatorNameLang":"en"}],"nameIdentifiers":[{}]}]},"item_4_source_id_9":{"attribute_name":"書誌レコードID","attribute_value_mlt":[{"subitem_source_identifier":"AN10115061","subitem_source_identifier_type":"NCID"}]},"item_4_textarea_12":{"attribute_name":"Notice","attribute_value_mlt":[{"subitem_textarea_value":"SIG Technical Reports are nonrefereed and hence may later appear in any journals, conferences, symposia, etc."}]},"item_resource_type":{"attribute_name":"資源タイプ","attribute_value_mlt":[{"resourceuri":"http://purl.org/coar/resource_type/c_18gh","resourcetype":"technical report"}]},"item_4_description_7":{"attribute_name":"論文抄録","attribute_value_mlt":[{"subitem_description":"文書があるトピックの持つ確率分布から生成されたと仮定し  その確率分布パラメータと文書のトピックへの帰属確率を求めるモデルに  ナイーブベイズ法を Polya 分布を用いてベイズ的に精密にとらえ直した混合ディリクレモデル(DM)があるが  この方法はトピック数を事前に与える必要があるという欠点があった.これに対し  本論文では可算無限個の混合比にディリクレ過程事前分布を与えることにより  データの複雑さに合わせて混合数を自動推定するディリクレ過程混合モデルによる方法を検討する. モデル選択により混合数を決定する方法と異なり  この方法は混合数の事後分布をパラメータと同時に推定し  期待値を取ることで予測を行うことができる.  実験の結果  必要な混合数の上限を推測することができ  特に小規模データに対しては性能がさらに上昇することがわかった.","subitem_description_type":"Other"}]},"item_4_description_8":{"attribute_name":"論文抄録(英)","attribute_value_mlt":[{"subitem_description":"This paper proposes a Dirichlet process mixture modeling approach to Dirichlet Mixtures (DM). Endowing a prior distribution on an infinite number of mixture components, this approach yields an appropriate number of components as well as their parameters at the same time. Experimental results on amino acid distributions and text corpora confirmed this effect and show comparative performance on large datasets and better performance on small datasets avoiding overfitting.","subitem_description_type":"Other"}]},"item_4_biblio_info_10":{"attribute_name":"書誌情報","attribute_value_mlt":[{"bibliographicPageEnd":"53","bibliographic_titles":[{"bibliographic_title":"情報処理学会研究報告自然言語処理（NL）"}],"bibliographicPageStart":"47","bibliographicIssueDates":{"bibliographicIssueDate":"2006-03-27","bibliographicIssueDateType":"Issued"},"bibliographicIssueNumber":"36(2006-NL-172)","bibliographicVolumeNumber":"2006"}]},"relation_version_is_last":true,"weko_creator_id":"1"},"created":"2025-01-18T23:13:18.031393+00:00","id":47941,"links":{}}