@techreport{oai:ipsj.ixsq.nii.ac.jp:00020254, author = {平出, 基一 and 田中, 栄一 and Motoichi, Hirade and Eiichi, Tanaka}, issue = {12(1994-DBS-101)}, month = {Jan}, note = {本文は拡張可能類名表記とB^+?木を用いて類似キーの検索ができるファイルの一構成法について書いている。拡張類名表記を用いると、従来の類名表記を用いたときに比べて、1つの類名表記に属するキーの数の分散が少なくなり、記憶利用率が良くなる。230,188個のキーで実験したところ、記憶利用率は約66%であった。長さ6の入力キーのとき、類似キー検索時のページ読み込み回数は約30.2回であった。紙数の都合で理論的検討の記述は省略したが、実験値は理論値とよく一致した。, In information retrieval there is a great demand for a file that can retrieve similar keys to an input key. However, only a few researches have been done for this kind of files. This paper proposes a file of this kind using extendible class name expressions and based on a B^+-tree. Therefore, the file can easily insert and delete keys without lowering its storage utilization. The experiment using 230,188 keys with length 1 縲鰀 16 shows the good performance of the file. That is, the file can retrieve, insert and delete keys with less disk access than an original B-tree. The number of page reads for searching similar keys is around 30.2 in case of an input key with length 6. The storage utilization of the file is about 66%. This value is very close to the average storage utilization of an original B-tree. The performance of this files is analyzed and the theoretical evaluation shows a good coincidence to the experimental value.}, title = {拡張可能類名表記を用いた類似キー検索ファイル}, year = {1995} }