A fuzzy-logic based search engine for a soybean shape database was dev
eloped on the WWW. It used to be difficult to evaluate crop shape char
acteristics quantitatively. Image analyzing technologies, however, can
now bring us several numerical parameters to describe crop shape and
we can develop database systems of such numerical shape characteristic
s combining some image data. To make a search on shape information in
such a database, one would rather prefer using naturally linguistic ke
ywords such as ''very fat and round'' than numerical keywords because
it is usually difficult to imagine appropriate values for what one wan
ts in terms of shape. Therefore, a search system based on fuzzy logic
with naturally linguistic and comparative keywords for shape informati
on was developed in this study. In the fuzzy-logic based search system
, a wide range of human subjectivity to the search results, caused mai
nly by the subjective recognition of each fuzzy keyword, can be distin
ct. Therefore, in order to adjust the system to such subjectivity, we
prepared several search options that is, two types of membership funct
ions, flexible fuzzy range and three types of definitions on fuzzy mem
bership logical products. One can try the system at http : //zoussooku
.narc. affrc.go.jp/similar to snino/soyshape/search2.html.