インターネットのサーチエンジンの評価尺度―ESL（expected search length）を使った検索実験

安形輝; Teru Agata; 野末道子; Michiko Nozue; 服部紀彦; Norihiko Hattori; 上田修一; Shuichi Ueda

doi:10.46895/lis.37.1

インターネットのサーチエンジンの評価尺度―ESL（expected search length）を使った検索実験A measure for evaluating search engines on the World Wide Web: Retrieval test with ESL (expected search length)

安形輝¹，野末道子²，服部紀彦³，上田修一³Teru Agata¹, Michiko Nozue², Norihiko Hattori³, Shuichi Ueda³

¹ 慶應義塾大学大学院文学研究科図書館・情報学専攻Graduate School of Library and Information Science, Keio University ◇ 〒108-8345 東京都港区三田二丁目15番45号 ◇ Mita 2-15-45, Minato-ku, Tokyo 108-8345, Japan

² 鉄道総合技術研究所輸送システム開発推進部Division of Transport Systems Development, Railway Technical Research Institute ◇ 〒185-8540 東京都国分寺市光町二丁目8番地38 ◇ Hikaricho 2-8-38, Kokubunji, Tokyo 185-8540, Japan

³ 慶鷹義塾大学文学部図書館・情報学科School of Library and Information Science, Keio University ◇ 〒108-8345 東京都港区三田二丁目15番45号 ◇ Mita 2-15-45, Minato-ku, Tokyo 108-8345, Japan

受付日：1998年1月27日Received: January 27, 1998

受理日：1998年9月8日Accepted: September 8, 1998

発行日：1998年9月30日Published: September 30, 1998

Search engines are a kind of information retrieval systems on the Internet. However, they cannot be evaluated with the measures commonly applied in traditional retrieval experiments, i.e., recall and precision, since recall cannot be calculated in the essentially unrestricted resources of the Internet. The Expected Search Length (ESL) proposed by Cooper in 1968 is a measure for evaluating information retrieval systems and has the possibility of taking the place of recall and precision. The ESL calculates the cost paid by a user, i.e., number of retrieved sites the user must look through before (s)he gets sufficient number of relevant sites. This article proposes a version of the ESL in order to adapt it to evaluate search engines. An experiment was carried out to compare the validity of the ESL with that of precision and recall. Eight search engines of the domestic and overseas were evaluated by precision and ESL. The result of this retrieval experiment shows the adaptability of ESL for the evaluation of the search engines.