@inproceedings{oai:ipsj.ixsq.nii.ac.jp:00240872,
 author = {須藤, 弘貴 and 内山, 智貴 and Hiroki, Sudo and Tomoki, Uchiyama},
 book = {コンピュータセキュリティシンポジウム2024論文集},
 month = {Oct},
 note = {本研究ではSQL Window関数を秘密計算上で効率的に実装する手法を提案する．SQL Window関数とは，データベースクエリの中でデータのグループごとかつ指定したフレームごとに集計や分析を行うための機能群である．秘密計算上でSQL Window関数を実現する既存研究では通信量がフレーム幅とレコード数の積に線形でありフレーム幅が定数でない(先頭行始まり/最終行終わり)場合に O(n2)通信量となるため実行時間が遅いという課題があった．本研究では，フレーム幅が定数でない場合にフレーム幅に依存せずO(n log n)通信量で代表的なSQL Window関数を実現するsemi-honest安全なプロトコルを提案する．さらに，提案プロトコルの実装について性能評価を行う．, In this study, we propose a method for efficiently implementing SQL Window functions in secure computation. SQL Window functions are a set of features in database queries that allow aggregation and analysis for each data group and specified frame. Existing research on implementing SQL Window functions in secure computation faces the challenge of high communication costs, which are linear with the product of the frame width and the number of records. When the frame width is not constant (e.g., starting from the first row or ending with the last row), the communication cost becomes O(n2), leading to slower execution times. In this study, we propose a semi-honest secure protocol that achieves the communication cost of O(n log n), independent of the frame width, even when the frame width is not constant, to implement representative SQL Window functions. Additionally, we conduct a performance evaluation of the proposed protocol's implementation.},
 pages = {939--946},
 publisher = {情報処理学会},
 title = {秘匿 SQL Window 関数の効率化},
 year = {2024}
}