{"links":{},"id":2003331,"metadata":{"_oai":{"id":"oai:ipsj.ixsq.nii.ac.jp:02003331","sets":["934:1022:11800:1749773804306"]},"path":["1749773804306"],"owner":"80578","recid":"2003331","title":["大規模言語モデルを用いた文書補強とリランキングによる統計データ検索"],"pubdate":{"attribute_name":"PubDate","attribute_value":"2025-07-29"},"_buckets":{"deposit":"87605540-b69a-4e76-b678-5becd5674738"},"_deposit":{"id":"2003331","pid":{"type":"depid","value":"2003331","revision_id":0},"owners":[80578],"status":"published","created_by":80578},"item_title":"大規模言語モデルを用いた文書補強とリランキングによる統計データ検索","author_link":[],"item_titles":{"attribute_name":"タイトル","attribute_value_mlt":[{"subitem_title":"大規模言語モデルを用いた文書補強とリランキングによる統計データ検索","subitem_title_language":"ja"},{"subitem_title":"Statistical Data Retrieval Using Document Augmentation and Re-ranking with Large Language Models","subitem_title_language":"en"}]},"item_keyword":{"attribute_name":"キーワード","attribute_value_mlt":[{"subitem_subject":"[研究論文] 統計データ検索，メタデータ，文書補強，大規模言語モデル，情報検索","subitem_subject_scheme":"Other"}]},"item_type_id":"3","publish_date":"2025-07-29","item_3_text_3":{"attribute_name":"著者所属","attribute_value_mlt":[{"subitem_text_value":"京都産業大学情報理工学部／現在，京都大学情報学研究科"},{"subitem_text_value":"京都産業大学情報理工学部"}]},"item_3_text_4":{"attribute_name":"著者所属(英)","attribute_value_mlt":[{"subitem_text_value":"Faculty of Information Science and Engineering, Kyoto Sangyo University / Presently with Graduate School of Informatics, Kyoto University","subitem_text_language":"en"},{"subitem_text_value":"Faculty of Information Science and Engineering, Kyoto Sangyo University","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/2003331/files/IPSJ-TOD1803004.pdf","label":"IPSJ-TOD1803004.pdf"},"date":[{"dateType":"Available","dateValue":"2027-07-29"}],"format":"application/pdf","billing":["billing_file"],"filename":"IPSJ-TOD1803004.pdf","filesize":[{"value":"6.5 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":"13"},{"tax":["include_tax"],"price":"0","billingrole":"39"},{"tax":["include_tax"],"price":"0","billingrole":"44"}],"accessrole":"open_date","version_id":"b6deb4f3-f026-405d-b9d1-8a0eea795c5f","displaytype":"detail","licensetype":"license_note","license_note":"Copyright (c) 2025 by the Information Processing Society of Japan"}]},"item_3_creator_5":{"attribute_name":"著者名","attribute_type":"creator","attribute_value_mlt":[{"creatorNames":[{"creatorName":"黒川,博生"}]},{"creatorNames":[{"creatorName":"宮森,恒"}]}]},"item_3_creator_6":{"attribute_name":"著者名(英)","attribute_type":"creator","attribute_value_mlt":[{"creatorNames":[{"creatorName":"Hiroki Kurokawa","creatorNameLang":"en"}]},{"creatorNames":[{"creatorName":"Hisashi Miyamori","creatorNameLang":"en"}]}]},"item_3_source_id_9":{"attribute_name":"書誌レコードID","attribute_value_mlt":[{"subitem_source_identifier":"AA11464847","subitem_source_identifier_type":"NCID"}]},"item_resource_type":{"attribute_name":"資源タイプ","attribute_value_mlt":[{"resourceuri":"http://purl.org/coar/resource_type/c_6501","resourcetype":"journal article"}]},"item_3_source_id_11":{"attribute_name":"ISSN","attribute_value_mlt":[{"subitem_source_identifier":"1882-7799","subitem_source_identifier_type":"ISSN"}]},"item_3_description_7":{"attribute_name":"論文抄録","attribute_value_mlt":[{"subitem_description":"統計データは，政府などが保有するオープンデータの一種であり，近年，社会問題となっているフェイクニュースに対処するための事実確認（ファクトチェック）への活用をはじめ，有効活用するためのアドホック検索基盤の重要性が高まっている．しかし，従来の統計データ検索では，データの大部分を占める数値の並びを活用できておらず，十分なランキング性能を達成できていない．そこで，本稿では大規模言語モデルを用いた文書補強とリランキングによる統計データのアドホック検索手法を提案する．提案手法では，まず統計データから抽出された見出し，行ヘッダ，列ヘッダ，値に基づき，その内容説明を大規模言語モデルで生成することでメタデータを補強した文書を作成する．次に，補強された文書を利用してランキングを行い，最後に大規模言語モデルを用いて意味内容の類似に基づくリランキングを行う．実験では，日本語のデータセットを用いて，提案手法と従来手法によるランキング性能を比較評価する．","subitem_description_type":"Other"}]},"item_3_description_8":{"attribute_name":"論文抄録(英)","attribute_value_mlt":[{"subitem_description":"Statistical data is a kind of open data provided by governments, statistics bureaus, etc. Recently, the importance of ad hoc retrieval infrastructure for effective use of such data has been increasing, including its use for fact-checking to deal with the social problem of fake news. However, conventional statistical data retrieval does not take advantage of the numerical sequence, which constitutes most of the data, and does not achieve sufficient ranking performance. Therefore, in this paper, we propose an ad hoc retrieval method for statistical data based on document augmentation and reranking using large language models. The proposed method first creates documents augmented with metadata based on headings, row names, column names, and values extracted from statistical data and then generates explanations of their contents using a large language model. Next, ranking is performed using the augmented documents, and finally, re-ranking is performed based on the similarity of semantic content using another large language model. In the experiment, we evaluate and compare the ranking performance of the proposed method and conventional method using Japanese dataset.","subitem_description_type":"Other"}]},"item_3_biblio_info_10":{"attribute_name":"書誌情報","attribute_value_mlt":[{"bibliographicPageEnd":"34","bibliographic_titles":[{"bibliographic_title":"情報処理学会論文誌データベース（TOD）"}],"bibliographicPageStart":"20","bibliographicIssueDates":{"bibliographicIssueDate":"2025-07-29","bibliographicIssueDateType":"Issued"},"bibliographicIssueNumber":"3","bibliographicVolumeNumber":"18"}]},"relation_version_is_last":true,"weko_creator_id":"80578"},"created":"2025-07-23T03:06:57.285583+00:00","updated":"2025-07-23T03:07:01.567739+00:00"}