{"metadata":{"_oai":{"id":"oai:ipsj.ixsq.nii.ac.jp:00214084","sets":["1164:4179:10535:10759"]},"path":["10759"],"owner":"44499","recid":"214084","title":["BERTに区間損失を加味した意見対象抽出"],"pubdate":{"attribute_name":"公開日","attribute_value":"2021-11-24"},"_buckets":{"deposit":"7a833755-153b-4c03-a687-09c5083ba248"},"_deposit":{"id":"214084","pid":{"type":"depid","value":"214084","revision_id":0},"owners":[44499],"status":"published","created_by":44499},"item_title":"BERTに区間損失を加味した意見対象抽出","author_link":["548568","548569"],"item_titles":{"attribute_name":"タイトル","attribute_value_mlt":[{"subitem_title":"BERTに区間損失を加味した意見対象抽出"},{"subitem_title":"Opinion Target Extraction using Span Loss to BERT","subitem_title_language":"en"}]},"item_keyword":{"attribute_name":"キーワード","attribute_value_mlt":[{"subitem_subject":"深層学習モデル","subitem_subject_scheme":"Other"}]},"item_type_id":"4","publish_date":"2021-11-24","item_4_text_3":{"attribute_name":"著者所属","attribute_value_mlt":[{"subitem_text_value":"NHK放送技術研究所スマートプロダクション研究部"}]},"item_4_text_4":{"attribute_name":"著者所属(英)","attribute_value_mlt":[{"subitem_text_value":"Japan Broadcasting Corp. Science & Technology Labs. Smart Production div","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/214084/files/IPSJ-NL21251021.pdf","label":"IPSJ-NL21251021.pdf"},"format":"application/pdf","billing":["billing_file"],"filename":"IPSJ-NL21251021.pdf","filesize":[{"value":"1.0 MB"}],"mimetype":"application/pdf","priceinfo":[{"tax":["include_tax"],"price":"0","billingrole":"23"},{"tax":["include_tax"],"price":"0","billingrole":"44"}],"accessrole":"open_login","version_id":"5574e534-a7c6-4db7-a507-7ec3ab9a7532","displaytype":"detail","licensetype":"license_note","license_note":"Copyright (c) 2021 by the Institute of Electronics, Information and Communication Engineers This SIG report is only available to those in membership of the SIG."}]},"item_4_creator_5":{"attribute_name":"著者名","attribute_type":"creator","attribute_value_mlt":[{"creatorNames":[{"creatorName":"小早川, 健"}],"nameIdentifiers":[{}]}]},"item_4_creator_6":{"attribute_name":"著者名(英)","attribute_type":"creator","attribute_value_mlt":[{"creatorNames":[{"creatorName":"Takeshi, S. Kobayakawa","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_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":"評判分析において，分析すべき意見文の中から，意見対象区間の抽出とその評価の両方を同時に行う必要がある場合があり，特に，意見対象のバリエーションが多い場合に有効な方法である．この種のタスクの場合，分析すべき意見文を全体としてポジ・ネガ判定するという文分類の問題としてではなく，分析すべき文のそれぞれの単語にラベルを付与するという系列ラベリング問題として扱う手法が有効である．しかし，従来の系列ラベリング問題では，区間の短いラベルに対しては高精度な抽出結果が得られているものの，評判分析のタスクにおいては精度が低い という問題点があった．この研究では，深層学習を用いた系列ラベルングのモデルに対して，ラベルとなりうる区間 をモデル化する手法を提案する．放送番組に関連したツイートを分析するという評判分析のタスクにおいて，このモデルの有効性を確認したので報告する．","subitem_description_type":"Other"}]},"item_4_description_8":{"attribute_name":"論文抄録(英)","attribute_value_mlt":[{"subitem_description":"Extracting both opinions and opinion-targes simultaneously from opinionated sentences is sometimes required for opinion mining, especially effective when opinion targets have very wide variety. In such case, dealing the whole sentence opinion classification is not suitable. Extracting opinions and opinion-targets as a sequence labeling problem is effective, instead. However, the problem suffers from the deterioration of span detection accuracy. We propose a model to tackle this problem, and report the effectiveness by experiments, where the testdata are tweets related to TV programs.","subitem_description_type":"Other"}]},"item_4_biblio_info_10":{"attribute_name":"書誌情報","attribute_value_mlt":[{"bibliographicPageEnd":"6","bibliographic_titles":[{"bibliographic_title":"研究報告自然言語処理（NL）"}],"bibliographicPageStart":"1","bibliographicIssueDates":{"bibliographicIssueDate":"2021-11-24","bibliographicIssueDateType":"Issued"},"bibliographicIssueNumber":"21","bibliographicVolumeNumber":"2021-NL-251"}]},"relation_version_is_last":true,"weko_creator_id":"44499"},"id":214084,"updated":"2025-01-19T16:54:29.832526+00:00","links":{},"created":"2025-01-19T01:14:55.157943+00:00"}