{"created":"2023-05-15T14:51:06.042726+00:00","id":4779,"links":{},"metadata":{"_buckets":{"deposit":"15f4b0ba-75bb-47d1-92aa-250fddeefb5f"},"_deposit":{"created_by":3,"id":"4779","owners":[3],"pid":{"revision_id":0,"type":"depid","value":"4779"},"status":"published"},"_oai":{"id":"oai:kokubunken.repo.nii.ac.jp:00004779","sets":["1:372:373:419"]},"author_link":["31207","31205","31204","31206"],"item_10002_biblio_info_7":{"attribute_name":"書誌情報","attribute_value_mlt":[{"bibliographicIssueDates":{"bibliographicIssueDate":"2023-03-24","bibliographicIssueDateType":"Issued"},"bibliographicIssueNumber":"49","bibliographicPageEnd":"226","bibliographicPageStart":"119","bibliographic_titles":[{"bibliographic_title":"国文学研究資料館紀要 文学研究篇"},{"bibliographic_title":"The Bulletin of The National Institure of Japanese Literature","bibliographic_titleLang":"en"}]}]},"item_10002_description_19":{"attribute_name":"フォーマット","attribute_value_mlt":[{"subitem_description":"pdf","subitem_description_type":"Other"}]},"item_10002_description_5":{"attribute_name":"抄録","attribute_value_mlt":[{"subitem_description":"　ディープランニングが進歩に伴い、人工智能（AI）による詩歌の創作研究が盛んに行われるようになった。著者の研究チームは最近深層学習による和歌生成モデル「和歌VT」を提案した。「和歌VT」は条件付き変分オートエンコーダーと自己注意メカニズムを組み合わせ、指定されたキーワードによって質の高い和歌を生成することができる。本論文ではDice 係数による類似和歌の検索方法を提案し、「和歌VT」モデルによって生成された３首のAI 和歌を例に挙げ、人間和歌との比較の観点から分析を行った。Dice 係数を計算するこの検索方法で類似和歌を客観的に順位付けし、膨大な和歌データから最も目標和歌と近い和歌を選び出すことができる。この和歌検索方法は研究者が和歌を検索する上での参考の一つとして提示することが出来ればと考えている。同方法によって、和歌生成モデル「和歌VT」が創作したAI 和歌と類似する人間和歌を選び出し、その文法や語彙の組み合わせを比較し、AI 和歌の特徴と独創性を分析した。分析結果により、「和歌VT」モデルは人間和歌の主題、文法、句の構造、及び用語を正確にマスターし、独創的な単語の組み合わせ方により斬新的な和歌を生成できることが認められた。\n　With the development of artificial intelligence(AI)technology, research on poetry creation has been receiving increasingly close attention. In this article, a Waka retrieval method that calulates the similarity index of Waka based on the Dice coefficient is proposed. The method can objectively rank similar Waka poems based on the similarity index, and provide an objective method by which researchers can select similar Wakas from a large number of Waka poems. Using this Waka retrieval method, we select human Waka poems that are similar to the AI Waka generated by the WakaVT model, and analyze the originality of AI Waka through comparing AI waka and human Waka in terms of the features of vocabulary matching. The analysis results show that the WakaVT model has accurately learned the words and grammar of human Waka, as well as the logical structure and causality between sentences, and can generate novel AI Waka using new word collocations and metaphors. This research is expected to be useful not only for the study of AI Waka but also for the study of human Waka poems.","subitem_description_type":"Abstract"}]},"item_10002_publisher_8":{"attribute_name":"出版者","attribute_value_mlt":[{"subitem_publisher":"国文学研究資料館"}]},"item_10002_source_id_9":{"attribute_name":"ISSN","attribute_value_mlt":[{"subitem_source_identifier":"2436-3316","subitem_source_identifier_type":"ISSN"}]},"item_creator":{"attribute_name":"著者","attribute_type":"creator","attribute_value_mlt":[{"creatorNames":[{"creatorName":"武石, 悠霞"},{"creatorName":"タケイシ, ユカ","creatorNameLang":"ja-Kana"}],"nameIdentifiers":[{}]},{"creatorNames":[{"creatorName":"金, 中"},{"creatorName":"ジン, チョン","creatorNameLang":"ja-Kana"}],"nameIdentifiers":[{}]},{"creatorNames":[{"creatorName":"TAKEISHI, Yuka","creatorNameLang":"en"}],"nameIdentifiers":[{}]},{"creatorNames":[{"creatorName":"JIN, Zhong","creatorNameLang":"en"}],"nameIdentifiers":[{}]}]},"item_files":{"attribute_name":"ファイル情報","attribute_type":"file","attribute_value_mlt":[{"accessrole":"open_date","date":[{"dateType":"Available","dateValue":"2023-04-05"}],"displaytype":"detail","filename":"KB4908.pdf","filesize":[{"value":"1.3 MB"}],"format":"application/pdf","licensetype":"license_11","mimetype":"application/pdf","url":{"label":"AI 和歌の文学的特徴分析――Dice係数による類似和歌検索方法の導入を合わせて――","url":"https://kokubunken.repo.nii.ac.jp/record/4779/files/KB4908.pdf"},"version_id":"29ef7d92-4aef-443b-943a-4a5730a373f4"}]},"item_keyword":{"attribute_name":"キーワード","attribute_value_mlt":[{"subitem_subject":"和歌生成","subitem_subject_scheme":"Other"},{"subitem_subject":"人工智能","subitem_subject_scheme":"Other"},{"subitem_subject":"Dice係数","subitem_subject_scheme":"Other"},{"subitem_subject":"Waka Generation","subitem_subject_language":"en","subitem_subject_scheme":"Other"},{"subitem_subject":"Artificial Intelligence","subitem_subject_language":"en","subitem_subject_scheme":"Other"},{"subitem_subject":"Dice Coefficient","subitem_subject_language":"en","subitem_subject_scheme":"Other"}]},"item_language":{"attribute_name":"言語","attribute_value_mlt":[{"subitem_language":"jpn"}]},"item_resource_type":{"attribute_name":"資源タイプ","attribute_value_mlt":[{"resourcetype":"departmental bulletin paper","resourceuri":"http://purl.org/coar/resource_type/c_6501"}]},"item_title":"AI 和歌の文学的特徴分析――Dice係数による類似和歌検索方法の導入を合わせて――","item_titles":{"attribute_name":"タイトル","attribute_value_mlt":[{"subitem_title":"AI 和歌の文学的特徴分析――Dice係数による類似和歌検索方法の導入を合わせて――"},{"subitem_title":"Literary Analysis of AI Waka: Introduction of a Similar Waka Retrieval Method Using the Dice Coefficient","subitem_title_language":"en"}]},"item_type_id":"10002","owner":"3","path":["419"],"pubdate":{"attribute_name":"公開日","attribute_value":"2023-04-05"},"publish_date":"2023-04-05","publish_status":"0","recid":"4779","relation_version_is_last":true,"title":["AI 和歌の文学的特徴分析――Dice係数による類似和歌検索方法の導入を合わせて――"],"weko_creator_id":"3","weko_shared_id":-1},"updated":"2023-05-15T15:05:03.617801+00:00"}