A system of computer-based instruction (CBI) authoring and delivery to
ols was developed in which generative CBI techniques were used to dyna
mically create instructional presentations from content databases duri
ng program execution. Three applications using these techniques are ex
amined with regard to trade-offs between ease of development and limit
ations imposed by program specificity: 1) A semantic network was used
to represent large bodies of terse facts to be memorized. The need for
authors to create large numbers of question screens was avoided by a
feature enabling questions and foils to be generated on-the-fly from d
atabase assertions as the student programs ran. 2) Technical vocabular
y training was accomplished with a generative approach in which an arr
ay of student learning activities were automatically generated from a
database minimally consisting of technical terms and their definitions
. 3) A more difficult program allowed greater interface and lesson con
trol flexibility for creating more sophisticated practice lessons in w
hich new instances were dynamically inserted into templates for questi
ons, anticipated answers and feedback. A common application theme emer
ging from field-testing the programs was their use in learning resourc
e centers to address specific learning objectives from larger courses.