A tabu search approach to generating test sheets for multiple assessment criteria

A computer skill certification test is one of the most important methods for evaluation of a student's computer ability. Since the tests are held frequently, one must efficiently and effectively compose test sheets from a large item bank containing over 10 000 test items. To certify student computer skills fairly, the composed test sheets must meet multiple assessment criteria such as the ratio of relevant concepts to be evaluated, the average discrimination degree, the difficulty degree, and the estimated testing time. One must allow significant time to compose an optimal test sheet from a large item bank by generating and testing each possible combination of test items. To cope with this problem, a Tabu search-based approach is proposed for more efficient composition of near-optimal test sheets from very large item banks, while meeting multiple assessment criteria. Based on the proposed approach, a computer-assisted testing system has been developed, and a series of experiments have been conducted to compare the efficiency and efficacy of this approach with other approaches. The experimental results show that the new approach is desirable for the composition of near-optimal test sheets from large item banks.