An efficient genetic algorithm for item selection strategy

Abstract

Genetic algorithm has achieved remarkable results in solving the problem of combinatorial optimization in the artificial intelligence area in recent years. However, efficient search for one reasonable best solution is still underway among the massive restricted conditions. This paper proposes an efficient genetic algorithm based on Item Response Theory (IRT) and enables effective search for optimal solution or near optimal solution under restricted conditions. By modify the evolutionary parameters and the goal function of the simple genetic algorithm, test quality is not only acceptable by test designers, but the practicability is also enhanced. The proposed method is a more effective tool for education assessment researcher as it successfully extends artificial intelligence-genetic algorithm applied in the educational assessment.