| Summary: | 碩士 === 國立臺中教育大學 === 教育測驗統計研究所 === 102 === This study investigated the different evolutionary algorithms in multi-dimensional item response theory test assembly of the results, which can efficiently and accurately meet the objectives of the test design. In the test assembly, how to test information function by creating objectives, in line with the selection of a set of test item in the item bank in order to achieve the objectives for the test evaluation important issues, current research mainly in the way of the unidimensional item response theory and focuses on the different evolutionary algorithms to reduce the amount of target error messages between the amount of test information, so set out to better meet the test objective test. Since the purpose of the test with the complexity and diversity of the trend in recent years to develop a multidimensional item response theory in order to better meet the needs of today's test framework.
This study proposes the theory test assembly studies Kalyanmoy Deb genetic algorithm and particle swarm algorithm in multi-dimensional item response theory. Simulation results show that the use of Kalyanmoy Deb genetic algorithm and particle swarm algorithm for the problem on the test assembly is a feasible approach, proposed method also has the stability of test assembly; addition, the distribution of the different exam, the target volume of messages and small error between the test information to improve its usefulness for educational assessment provides an effective tool for researchers.
|
|---|
