WWW OPAC has become one of standard services at university libraries in Jap
an. A large number of studies have been carried out for the evaluation of O
PACs. What seem to be lacking, however, are those from the actual users' po
int of view. In this research, we examine major 15 WWW OPAC systems from su
ch search terms as supposed likely to be used by novice users. Three cases
are analyzed, each searching a specific book or author from terms extracted
from it and their variants. The first case is to search a Japanese book tr
anslated from English from 9 terms lex, full Japanese title, one keyword fr
om title,'kanji' of Japanese translator, English spell of author, and so on
). The second case is to search a book with long title from 5 terms (full t
itle including subtitle, main title only, and the variants). The third case
is to search all books by an author, whose 'yomi' is the same with more th
an 9 persons.
The major results are as follows:
(1) Only three systems are successful in searching from all the terms in th
e first case.
(2) Some systems are very ineffective, can be searched only from few terms.
(3) The same outcomes cannot be got even from the same search terms.
These problems seem to occur under such situations as:
(1) A standard architecture and indexing system of OPAC database have not y
et been established.
(2) Standard OPAC searching procedure has not been developed.
(3) Current OPAC is a mixture of traditional cataloguing rules and online r
etrieval system.
It is necessary to develop an original architecture and standard searching
procedure far OPAC based on a new user model.