Library and Information Science Society Home → English page

三田図書館・情報学会誌論文(論文ID LIS012155)

著者
吉村典夫五十嵐瞳
和文タイトル
検索式のインデクスの数と検索される課題数との相関:研究課題検索システム(REGISTER)における実験
英文タイトル
Correlation between the Number of Codes Used in Searching Formulas and the Number of Current Research Projects Retrieved
掲載号・頁
No.12, p.155-175
発行日
1974-10-31
英文抄録

REGISTER is a computer-based system for retrieval of research projects currently conducted in Japan covering all fields of science and technology, has been developed since 1970, and started its actual operation in 1972. Over and beyond the operation of the system, the Division of System Development, Japan Science Foundation, performs various investigations needed for improving the system and developing a new system called RECRAS aimed at retrieving the same kind of information on current research projects in the fields of agriculture, forestry and fishery in Japan.

The main subject of the present paper is one of experimental studies executed by the said Division concerning the quantitative relation between the number of codes in searching formulas and the number of research projects retrieved. For understanding the system, REGISTER, however, the writers describe its outline in the first part of the paper, and then, deal with the experimental study.

The data base of REGISTER is formed by assigning to research project classification codes according to a decimal classification scheme of four digits, CST, using a multi-aspect post-coordinating method. The retrieval is conducted by using and-logical expressions consisting of these classification codes as searching formulas.

Concerning the quantitative relation between the number of codes in searching formulas and the number of research projects retrieved by matching to the searching formulas, the writers assume that “the logarithm of the number of research projects retrieved becomes a linear function of the number of codes used in searching formulas, and the regression coefficient is negative” or our hypothesis is formularized as follows:

種別
原著論文