A library of templates for exercise construction in an adaptive assessment system

Conventional computer testing systems usually offer items (questions) where examinees can only select one or more answers. In these systems, there is a fixed number of items to be administered to students. This implies that all students must answer the same number of questions, independently of their knowledge level. By contrast, adaptive testing systems are able to make more accurate predictions of student knowledge level with shorter tests, by choosing the most adequate item to ask next, depending on the current estimation of student knowledge level. In order to be reliable, adaptive testing requires a well-founded underlying theory. The theory mainly used is the Item Response Theory. On the other hand, items that usually appear in adaptive tests have a simple and maybe boring format: a stem and a set of answers. In this paper, a library of templates for the automatic construction of more sophisticated items is presented. These types of items are used in SIETTE, an adaptive web-based system for knowledge assessment by means of tests. This system also provides the capability of generating isomorphic items. With the same mechanism used in the construction of this library, SIETTE has been provided with the capability of restricting the time consumed by students while taking tests.