{"id":218700,"created":"2025-01-19T01:19:03.247117+00:00","metadata":{"_oai":{"id":"oai:ipsj.ixsq.nii.ac.jp:00218700","sets":["1164:4179:10952:10953"]},"path":["10953"],"owner":"44499","recid":"218700","title":["フレーズアライメントと文構造に基づくデータ拡張を用いた頑健な自然言語生成"],"pubdate":{"attribute_name":"公開日","attribute_value":"2022-06-22"},"_buckets":{"deposit":"bf5e2101-3cbb-4f35-a6a2-2c6a3cfe91f4"},"_deposit":{"id":"218700","pid":{"type":"depid","value":"218700","revision_id":0},"owners":[44499],"status":"published","created_by":44499},"item_title":"フレーズアライメントと文構造に基づくデータ拡張を用いた頑健な自然言語生成","author_link":["569339","569337","569340","569338"],"item_titles":{"attribute_name":"タイトル","attribute_value_mlt":[{"subitem_title":"フレーズアライメントと文構造に基づくデータ拡張を用いた頑健な自然言語生成"}]},"item_keyword":{"attribute_name":"キーワード","attribute_value_mlt":[{"subitem_subject":"対話","subitem_subject_scheme":"Other"}]},"item_type_id":"4","publish_date":"2022-06-22","item_4_text_3":{"attribute_name":"著者所属","attribute_value_mlt":[{"subitem_text_value":"京都大学大学院情報学研究科"},{"subitem_text_value":"理化学研究所ガーディアンロボットプロジェクト"},{"subitem_text_value":"京都大学大学院情報学研究科"},{"subitem_text_value":"理化学研究所ガーディアンロボットプロジェクト"}]},"item_4_text_4":{"attribute_name":"著者所属(英)","attribute_value_mlt":[{"subitem_text_value":"Graduate school of informatics, Kyoto University","subitem_text_language":"en"},{"subitem_text_value":"GRP, RIKEN","subitem_text_language":"en"},{"subitem_text_value":"Graduate school of informatics, Kyoto University","subitem_text_language":"en"},{"subitem_text_value":"GRP, RIKEN","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/218700/files/IPSJ-NL22252010.pdf","label":"IPSJ-NL22252010.pdf"},"date":[{"dateType":"Available","dateValue":"2024-06-22"}],"format":"application/pdf","billing":["billing_file"],"filename":"IPSJ-NL22252010.pdf","filesize":[{"value":"3.9 MB"}],"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":"c4878fbf-5642-4d66-80eb-d9f9edc5bdba","displaytype":"detail","licensetype":"license_note","license_note":"Copyright (c) 2022 by the Information Processing Society of Japan"}]},"item_4_creator_5":{"attribute_name":"著者名","attribute_type":"creator","attribute_value_mlt":[{"creatorNames":[{"creatorName":"山本, 賢太"}],"nameIdentifiers":[{}]},{"creatorNames":[{"creatorName":"河野, 誠也"}],"nameIdentifiers":[{}]},{"creatorNames":[{"creatorName":"河原, 達也"}],"nameIdentifiers":[{}]},{"creatorNames":[{"creatorName":"吉野, 幸一郎"}],"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_source_id_11":{"attribute_name":"ISSN","attribute_value_mlt":[{"subitem_source_identifier":"2188-8779","subitem_source_identifier_type":"ISSN"}]},"item_4_description_7":{"attribute_name":"論文抄録","attribute_value_mlt":[{"subitem_description":"自然言語生成タスクは意味表現を入力として対応するテキスト（発話）を生成するタスクで，対話システムにおける重要なタスクのひとつである．近年は，ニューラルネットワークを用いた手法により，自然な言語生成が実現されている．ニューラルネットワークを用いた言語生成は学習データへの依存性が大きく，しばしば与えた意味表現にない情報を生成文に含めてしまう過生成の問題が生じる．そこで，本研究では，言語生成タスクのデータセットから意味表現を削除した学習データのバリエーションを作成するデータ拡張手法により，より頑健な自然言語生成を実現する．具体的には，学習データに含まれる意味表現に対して一部のスロットの削除を行い，この意味表現に対応して学習データ中の文を編集することで，既存の学習データに含まれない意味表現の組み合わせに対応したデータの拡張を行う．この対応取得のため，フレーズアライメントや注意機構の重みを用いる．また，文の編集を行う際にもとの文の構造を考慮する．実験では，提案法により過生成の問題を抑制しつつ，自然性も高い言語生成を行うことができることが確認された．","subitem_description_type":"Other"}]},"item_4_biblio_info_10":{"attribute_name":"書誌情報","attribute_value_mlt":[{"bibliographicPageEnd":"8","bibliographic_titles":[{"bibliographic_title":"研究報告自然言語処理（NL）"}],"bibliographicPageStart":"1","bibliographicIssueDates":{"bibliographicIssueDate":"2022-06-22","bibliographicIssueDateType":"Issued"},"bibliographicIssueNumber":"10","bibliographicVolumeNumber":"2022-NL-252"}]},"relation_version_is_last":true,"weko_creator_id":"44499"},"updated":"2025-01-19T15:04:35.332949+00:00","links":{}}