On the development of a computer-assisted testing system with genetic test sheet-generating approach

Abstract

Since the last decade, computer-assisted testing has proven to be an efficient and effective way to evaluating students' learning status such that proper tutoring strategies can be adopted to improve-their learning performance. A good test will not only help the instructor evaluate the learning status of the students, but also facilitate the diagnosis of the problems embedded in the students' learning process. One of the most important and challenging issues in conducting a good test is the construction of test sheets that can meet various assessment requirements. A previous study has indicated that selecting test items to best fit multiple assessment requirements can be formulated as a mixed integer programming model. The problem is known to be NP-hard in the literature and, hence, computational challenges hinder the development of efficient solution methods. As a sequel, we instead seek quality approximate solutions in a reasonable time. Two approximation methods based upon a genetic approach are developed. Statistics from a series of computational experiments indicate that our approach is able to efficiently generate near-optimal combinations of test items that satisfy the specified requirements or constraints.