An improved personalized genetic algorithm incorporated item distribution for test sheet assembling

In recent years, computer-based testing has become an effective approach to evaluate students' learning level. In our previous work, a personalized genetic algorithm (PGA) for test sheet assembling was proposed. In this paper, an improved personalized genetic algorithm named PGAC which makes an improvement in the crossover process of PGA is presented. Considering item distribution, an improved algorithm incorporated item distribution (IGAID) based on PGAC is presented to assemble simulation test sheets which have good item distribution in knowledge hierarchy for each student. Experiments and comparison with random assembling algorithm and GA are conducted. The results show that PGAC supports effectively assembling a test sheet with more non-mastered items for different students, and IGAID is capable of effectively constructing simulation test sheets with good item distribution in knowledge hierarchy for each student.